This manuscript was automatically generated from dhimmel/rephetio-manuscript@7b5b0e6 on October 3, 2017.
Daniel S. Himmelstein
0000-0002-3012-7446 · dhimmel · dhimmel
Program in Biological & Medical Informatics, University of California, San Francisco; Department of Systems Pharmacology & Translational Therapeutics, University of Pennsylvania
Department of Neurology, University of California, San Francisco
Sabrina L. Chen
Department of Neurology, University of California, San Francisco; Johns Hopkins University
Department of Neurology, University of California, San Francisco
Department of Neurology, University of California, San Francisco; Center for Neuroengineering and Therapeutics, University of Pennsylvania
The ability to computationally predict whether a compound treats a disease would improve the economy and success rate of drug approval. This study describes Project Rephetio to systematically model drug efficacy based on 755 existing treatments. First, we constructed Hetionet (neo4j.het.io), an integrative network encoding knowledge from millions of biomedical studies. Hetionet v1.0 consists of 47,031 nodes of 11 types and 2,250,197 relationships of 24 types. Data was integrated from 29 public resources to connect compounds, diseases, genes, anatomies, pathways, biological processes, molecular functions, cellular components, pharmacologic classes, side effects, and symptoms. Next, we identified network patterns that distinguish treatments from non-treatments. Then we predicted the probability of treatment for 209,168 compound–disease pairs (het.io/repurpose). Our predictions validated on two external sets of treatment and provided pharmacological insights on epilepsy, suggesting they will help prioritize drug repurposing candidates. This study was entirely open and received realtime feedback from 40 community members.
The cost of developing a new therapeutic drug has been estimated at 1.4 billion dollars , the process typically takes 15 years from lead compound to market , and the likelihood of success is stunningly low . Strikingly, the costs have been doubling every 9 years since 1970, a sort of inverse Moore’s law, which is far from an optimal strategy from both a business and public health perspective . Drug repurposing — identifying novel uses for existing therapeutics — can drastically reduce the duration, failure rates, and costs of approval . These benefits stem from the rich preexisting information on approved drugs, including extensive toxicology profiling performed during development, preclinical models, clinical trials, and postmarketing surveillance.
Drug repurposing is poised to become more efficient as mining of electronic health records (EHRs) to retrospectively assess the effect of drugs gains feasibility [6–9]. However, systematic approaches to repurpose drugs based on mining EHRs alone will likely lack power due to multiple testing. Similar to the approach followed to increase the power of genome-wide association studies (GWAS) [10,11], integration of biological knowledge to prioritize drug repurposing will help overcome limited EHR sample size and data quality.
In addition to repurposing, several other paradigm shifts in drug development have been proposed to improve efficiency. Since small molecules tend to bind to many targets, polypharmacology aims to find synergy in the multiple effects of a drug . Network pharmacology assumes diseases consist of a multitude of molecular alterations resulting in a robust disease state. Network pharmacology seeks to uncover multiple points of intervention into a specific pathophysiological state that together rehabilitate an otherwise resilient disease process [13,14]. Although target-centric drug discovery has dominated the field for decades, phenotypic screens have more recently resulted in a comparatively higher number of first-in-class small molecules . Recent technological advances have enabled a new paradigm in which mid- to high-throughput assessment of intermediate phenotypes, such as the molecular response to drugs, is replacing the classic target discovery approach [16–18]. Furthermore, integration of multiple channels of evidence, particularly diverse types of data, can overcome the limitations and weak performance inherent to data of a single domain . Modern computational approaches offer a convenient platform to tie these developments together as the reduced cost and increased velocity of in silico experimentation massively lowers the barriers to entry and price of failure [20,21].
Hetnets (short for heterogeneous networks) are networks with multiple types of nodes and relationships. They offer an intuitive, versatile, and powerful structure for data integration by aggregating graphs for each relationship type onto common nodes. In this study, we developed a hetnet (Hetionet v1.0) by integrating knowledge and experimental findings from decades of biomedical research spanning millions of publications. We adapted an algorithm originally developed for social network analysis and applied it to Hetionet v1.0 to identify patterns of efficacy and predict new uses for drugs. The algorithm performs edge prediction through a machine learning framework that accommodates the breadth and depth of information contained in Hetionet v1.0 [22,23]. Our approach represents an in silico implementation of network pharmacology that natively incorporates polypharmacology and high-throughput phenotypic screening.
One fundamental characteristic of our method is that it learns and evaluates itself on existing medical indications (i.e. a “gold standard”). Next, we introduce previous approaches that also performed comprehensive evaluation on existing treatments. A 2011 study, named PREDICT, compiled 1,933 treatments between 593 drugs and 313 diseases . Starting from the premise that similar drugs treat similar diseases, PREDICT trained a classifier that incorporates 5 types of drug-drug and 2 types of disease-disease similarity. A 2014 study compiled 890 treatments between 152 drugs and 145 diseases with transcriptional signatures . The authors found that compounds triggering an opposing transcriptional response to the disease were more likely to be treatments, although this effect was weak and limited to cancers. A 2016 study compiled 402 treatments between 238 drugs and 78 diseases and used a single proximity score — the average shortest path distance between a drug’s targets and disease’s associated proteins on the interactome — as a classifier .
We build on these successes by creating a framework for incorporating the effects of any biological relationship into the prediction of whether a drug treats a disease. By doing this, we were able to capture a multitude of effects that have been suggested as influential for drug repurposing including drug-drug similarity [24,27], disease-disease similarity [24,28], transcriptional signatures [17,18,25,29,30], protein interactions , genetic association [31,32], drug side effects [33,34], disease symptoms , and molecular pathways . Our ability to create such an integrative model of drug efficacy relies on the hetnet data structure to unite diverse information. On Hetionet v1.0, our algorithm learns which types of compound–disease paths discriminate treatments from non-treatments in order to predict the probability that a compound treats a disease.
We refer to this study as Project Rephetio (pronounced as rep-het-ee-oh). Both Rephetio and Hetionet are portmanteaus combining the words repurpose, heterogeneous, and network with the URL het.io.
We obtained and integrated data from 29 publicly available resources to create Hetionet v1.0 (Figure 1). The hetnet contains 47,031 nodes of 11 types (Table 1) and 2,250,197 relationships of 24 types (Table 2). The nodes consist of 1,552 small molecule compounds and 137 complex diseases, as well as genes, anatomies, pathways, biological processes, molecular functions, cellular components, perturbations, pharmacologic classes, drug side effects, and disease symptoms. The edges represent relationships between these nodes and encompass the collective knowledge produced by millions of studies over the last half century.
For example, Compound–binds–Gene edges represent when a compound binds to a protein encoded by a gene. This information has been extracted from the literature by human curators and compiled into databases such as DrugBank, ChEMBL, DrugCentral, and BindingDB. We combined these databases to create 11,571 binding edges between 1,389 compounds and 1,689 genes. These edges were compiled from 10,646 distinct publications, which Hetionet binding edges reference as an attribute. Binding edges represent a comprehensive catalog constructed from low throughput experimentation. However, we also integrated findings from high throughput technologies — many of which have only recently become available. For example, we generated consensus transcriptional signatures for compounds in LINCS L1000 and diseases in STARGEO.
While Hetionet v1.0 is ideally suited for drug repurposing, the network has broader biological applicability. For example, we have prototyped queries for a) identifying drugs that target a specific pathway, b) identifying biological processes involved in a specific disease, c) identifying the drug targets responsible for causing a specific side effect, and d) identifying anatomies with transcriptional relevance for a specific disease . Each of these queries was simple to write and took less than a second to run on our publicly available Hetionet Browser. While it is possible that existing services provide much of the aforementioned functionality, they offer less versatility. Hetionet differentiates itself in its ability to flexibly query across multiple domains of information. As a proof of concept, we enhanced the biological process query (b), which identified processes that were enriched for disease-associated genes, using multiple sclerosis (MS) as an example disease. The verbose Cypher code for this query is shown below:
MATCH path = // Specify the type of path to match (n0:Disease)-[e1:ASSOCIATES_DaG]-(n1:Gene)-[:INTERACTS_GiG]- (n2:Gene)-[:PARTICIPATES_GpBP]-(n3:BiologicalProcess) WHERE // Specify the source and target nodes n0.name = 'multiple sclerosis' AND n3.name = 'retina layer formation' // Require GWAS support for the Disease-associates-Gene relationship AND 'GWAS Catalog' in e1.sources // Require the interacting gene to be upregulated in a relevant tissue AND exists((n0)-[:LOCALIZES_DlA]-(:Anatomy)-[:UPREGULATES_AuG]-(n2)) RETURN path
The query above identifies genes that interact with MS GWAS-genes. However, interacting genes are discarded unless they are upregulated in an MS-related anatomy (i.e. anatomical structure, e.g. organ or tissue). Then relevant biological processes are identified. Thus, this single query spans 4 node and 5 relationship types.
The integrative potential of Hetionet v1.0 is reflected by its connectivity. Among the 11 metanodes, there are 66 possible source–target pairs. However, only 11 of them have at least one direct connection. In contrast, for paths of length 2, 50 pairs have connectivity (paths types that start on the source node type and end on the target node type, see Figure 1C). At length 3, all 66 pairs are connected. At length 4, the source–target pair with the fewest types of connectivity (Side Effect to Symptom) has 13 metapaths, while the pair with the most connectivity types (Gene to Gene) has 3,542 pairs. This high level of connectivity across a diversity of biomedical entities forms the foundation for automated translation of knowledge into biomedical insight.
Hetionet v1.0 is accessible via a Neo4j Browser at https://neo4j.het.io. This public Neo4j instance provides users an installation-free method to query and visualize the network. The Browser contains a tutorial guide as well as guides with the details of each Project Rephetio prediction. Hetionet v1.0 is also available for download in JSON, Neo4j, and TSV formats . The JSON and Neo4j database formats include node and edge properties — such as URLs, source and license information, and confidence scores — and are thus recommended.
One aim of Project Rephetio was to systematically evaluate how drugs exert their therapeutic potential. To address this question, we compiled a gold standard of 755 disease-modifying indications, which form the Compound–treats–Disease edges in Hetionet v1.0. Next, we identified types of paths (metapaths) that occurred more frequently between treatments than non-treatments (any compound–disease pair that is not a treatment). The advantage of this approach is that metapaths naturally correspond to mechanisms of pharmacological efficacy. For example, the Compound–binds–Gene–associates–Disease (CbGaD) metapath identifies when a drug binds to a protein corresponding to a gene involved in the disease.
We evaluated all 1,206 metapaths that traverse from compound to disease and have length of 2–4 (Figure 2A). To control for the different degrees of nodes, we used the degree-weighted path count (DWPC, see Methods) — which downweights paths going through highly-connected nodes  — to assess path prevalence. In addition, we compared the performance of each metapath to a baseline computed from permuted networks. Hetnet permutation preserves node degree while eliminating edge specificity, allowing us to isolate the portion of unpermuted metapath performance resulting from actual network paths. We refer to the permutation-adjusted performance measure as Δ AUROC. A positive Δ AUROC indicates that paths of the given type tended to occur more frequently between treatments than non-treatments, after accounting for different levels of connectivity (node degrees) in the hetnet. In general terms, Δ AUROC assesses whether paths of a given type were informative of drug efficacy.
Overall, 709 of the 1,206 metapaths exhibited a statistically significant Δ AUROC at a false discovery rate cutoff of 5%. These 709 metapaths included all 24 metaedges, suggesting that each type of relationship we integrated provided at least some therapeutic utility. However, not all metaedges were equally present in significant metapaths: 259 significant metapaths included a Compound–binds–Gene metaedge, whereas only 4 included a Gene–participates–Cellular Component metaedge. Table 3 lists the predictiveness of several metapaths of interest. Refer to the Discussion for our interpretation of these findings.
We implemented a machine learning approach to translate the network connectivity between a compound and a disease into a probability of treatment [40,41]. The approach relies on the 755 known treatments as positives and 29,044 non-treatments as negatives to train a logistic regression model. Note that 179,369 non-treatments were omitted as negative training observations because they had a prior probability of treatment equal to zero (see Methods). The features consisted of a prior probability of treatment, node degrees for 14 metaedges, and DWPCs for 123 metapaths that were well suited for modeling. A cross-validated elastic net was used to minimize overfitting, yielding a model with 31 features (Figure 2B). The DWPC features with negative coefficients appear to be included as node-degree-capturing covariates, i.e. they reflect the general connectivity of the compound and disease rather than specific paths between them. However, the 11 DWPC features with non-negligible positive coefficients represent the most salient types of connectivity for systematically modeling drug efficacy. See the metapaths with positive coefficients in Table 3 for unabbreviated names. As an example, the CcSEcCtD feature assesses whether the compound causes the same side effects as compounds that treat the disease. Alternatively, the CbGeAlD feature assesses whether the compound binds to genes that are expressed in the anatomies affected by the disease.
We applied this model to predict the probability of treatment between each of 1,538 connected compounds and each of 136 connected diseases, resulting in predictions for 209,168 compound–disease pairs , available at http://het.io/repurpose/. The 755 known disease-modifying indications were highly ranked (AUROC = 97.4%, Figure 3). The predictions also successfully prioritized two external validation sets: novel indications from DrugCentral (AUROC = 85.5%) and novel indications in clinical trial (AUROC = 70.0%). Together, these findings indicate that Project Rephetio has the ability to recognize efficacious compound–disease pairs.
Predictions were scaled to the overall prevalence of treatments (0.36%). Hence a compound–disease pair that received a prediction of 1% represents a 2-fold enrichment over the null probability. Of the 3,980 predictions with a probability exceeding 1%, 586 corresponded to known disease-modifying indications, leaving 3,394 repurposing candidates. For a given compound or disease, we provide the percentile rank of each prediction. Therefore, users can assess whether a given prediction is a top prediction for the compound or disease. In addition, our table-based prediction browser links to a custom guide for each prediction, which displays in the Neo4j Hetionet Browser. Each guide includes a query to display the top paths supporting the prediction and lists clinical trials investigating the indication.
There are currently two FDA-approved medications for smoking cessation (varenicline and bupropion) that are not nicotine replacement therapies. PharmacotherapyDB v1.0 lists varenicline as a disease-modifying indication and nicotine itself as a symptomatic indication for nicotine dependence, but is missing bupropion. Bupropion was first approved for depression in 1985. Owing to the serendipitous observation that it decreased smoking in depressed patients taking this drug, Bupropion was approved for smoking cessation in 1997 . Therefore we looked whether Project Rephetio could have predicted this repurposing. Bupropion was the 9th best prediction for nicotine dependence (99.5th percentile) with a probability 2.50-fold greater than the null. Figure 4 shows the top paths supporting the repurposing of bupropion.
Atop the nicotine dependence predictions were nicotine (10.97-fold over null), cytisine (10.58-fold), and galantamine (9.50-fold). Cytisine is widely used in Eastern Europe for smoking cessation due to its availability at a fraction of the cost of other pharmaceutical options . In the last half decade, large scale clinical trials have confirmed cytisine’s efficacy [49,50]. Galantamine, an approved Alzheimer’s treatment, is currently in Phase 2 trial for smoking cessation and is showing promising results . In summary, nicotine dependence illustrates Project Rephetio’s ability to predict efficacious treatments and prioritize historic and contemporary repurposing opportunities.
Several factors make epilepsy an interesting disease for evaluating repurposing predictions . Antiepileptic drugs work by increasing the seizure threshold — the amount of electric stimulation that is required to induce seizure. The effect of a drug on the seizure threshold can be cheaply and reliably tested in rodent models. As a result, the viability of most approved drugs in treating epilepsy is known.
We focused our evaluation on the top 100 scoring compounds — referred to as the epilepsy predictions in this section — after discarding a single combination drug. We classified each compound as anti-ictogenic (seizure suppressing), unknown (no established effect on the seizure threshold), or ictogenic (seizure generating) according to medical literature . Of the top 100 epilepsy predictions, 77 were anti-ictogenic, 8 were unknown, and 15 were ictogenic (Figure 5A). Notably, the predictions contained 23 of the 25 disease-modifying antiepileptics in PharamcotherapyDB v1.0.
Many of the 77 anti-ictogenic compounds were not first-line antiepileptic drugs. Instead, they were used as ancillary drugs in the treatment of status epilepticus. For example, we predicted four halogenated ethers, two of which (isoflurane and desflurane) are used clinically to treat life-threatening seizures that persist despite treatment . As inhaled anesthetics, these compounds are not appropriate as daily epilepsy medications, but are feasible for refractory status epilepticus where patients are intubated.
Given this high precision (77%), the 8 compounds of unknown effect are promising repurposing candidates. For example, acamprosate — whose top prediction was epilepsy — is a taurine analog that promotes alcohol abstinence. Support for this repurposing arose from acamprosate’s inhibition of the glutamate receptor and positive modulation of the GABAᴀ receptor (Figure 5C). If effective against epilepsy, acamprosate could serve a dual benefit for recovering alcoholics who experience seizures from alcohol withdrawal.
While certain classes of compounds were highly represented in our epilepsy predictions, such benzodiazepines and barbiturates, there was also considerable diversity . The 100 predicted compounds encompassed 26 third-level ATC codes , such as antiarrhythmics (quinidine, classified as anti-ictogenic) and urologicals (phenazopyridine, classified as unknown). Furthermore, 25 of the compounds were chemically distinct, i.e. they did not resemble any of the other epilepsy predictions (Figure 5B).
Next, we investigated which components of Hetionet contributed to the epilepsy predictions . In total, 392,956 paths of 12 types supported the predictions. Using several different methods for grouping paths, we were able to quantify the aggregate biological evidence. Our algorithm primarily drew on two aspects of epilepsy: its known treatments (76% of the total support) and its genetic associations (22% of support). In contrast, our algorithm drew heavily on several aspects of the predicted compounds: their targeted genes (44%), their chemically similar compounds (30%), their pharmacologic classes, their palliative indications (5%), and their side effects (4%).
Specifically, 266,192 supporting paths originated with a Compound–binds–Gene relationship. Aggregating support by these genes shows the extent that 121 different drug targets contributed to the predictions . In order of importance, the predictions targeted GABAᴀ receptors (15.3% of total support), cytochrome P450 enzymes (5.6%), the sodium channel (4.6%), glutamate receptors (3.8%), the calcium channel (2.7%), carbonic anhydrases (2.5%), cholinergic receptors (2.1%) and the potassium channel (1.4%). Besides cytochrome P450, which primarily influences pharmacokinetics , our method detected and leveraged bonafide anti-ictogenic mechanisms . Figure 5C shows drug target contributions per compound and illustrates the considerable mechanistic diversity among the predictions.
Also notable are the 15 ictogenic compounds in our top 100 predictions. Nine of the ictogenic compounds share a tricyclic structure (Figure 5B), five of which are tricyclic antidepressants. While the ictogenic mechanisms of these antidepressants are still unclear , Figure 5C suggests their anticholinergic effects may be responsible , in accordance with previous theories .
We also ranked the contribution of the 1,137 side effects that supported the epilepsy predictions through 117,720 CcSEcCtD paths. The top five side effects — ataxia (0.069% of total support), nystagmus (0.049%), diplopia (0.045%), somnolence (0.044%), and vomiting (0.043%) — reflect established adverse effects of antiepileptic drugs [61–65]. In summary, our method simultaneously identified the hallmark side effects of antiepileptic drugs while incorporating this knowledge to prioritize 1,538 compounds for anti-ictogenic activity.
We created Hetionet v1.0 by integrating 29 resources into a single data structure — the hetnet. Consisting of 11 types of nodes and 24 types of relationships, Hetionet v1.0 brings more types of information together than previous leading-studies in biological data integration . Moreover, we strove to create a reusable, extensible, and property-rich network. While all of the resources we include are publicly available, their integration was a time-intensive undertaking and required careful consideration of legal barriers to data reuse. Hetionet allows researchers to begin answering integrative questions without having to first spend months processing data.
Our public Neo4j instance allows users to immediately interact with Hetionet. Through the Cypher language, users can perform highly specialized graph queries with only a few lines of code. Queries can be executed in the web browser or programmatically from a language with a Neo4j driver. For users that are unfamiliar with Cypher, we include several example queries in a Browser guide. In contrast to traditional REST APIs, our public Neo4j instance provides users with maximal flexibility to construct custom queries by exposing the underlying database.
As data has grown more plentiful and diverse, so has the applicability of hetnets. Unfortunately, network science has been naturally fragmented by discipline resulting in relatively slow progress in integrating heterogeneous data. A 2014 analysis identified 78 studies using multilayer networks — a superset of hetnets (heterogeneous information networks) with the potential for additional dimensions, such as time. However, the studies relied on 26 different terms, 9 of which had multiple definitions [67,68]. Nonetheless, core infrastructure and algorithms for hetnets are emerging. Compared to the existing mathematical frameworks for multilayer networks that must deal with layers other than type (such as the aspect of time) , the primary obligation of hetnet algorithms is to be type aware. One goal of our project has been to unite hetnet research across disciplines. We approached this goal by making Project Rephetio entirely available online and inviting community feedback throughout the process .
Integrating every resource into a single interconnected data structure allowed us to assess systematic mechanisms of drug efficacy. Using the max performing metapath to assess the pharmacological utility of a metaedge (Figure 2A), we can divide our relationships into tiers of informativeness. The top tier consists of the types of information traditionally considered by pharmacology: Compound–treats–Disease, Pharmacologic Class–includes–Compound, Compound–resembles–Compound, Disease–resembles–Disease, and Compound–binds–Gene. The upper-middle tier consists of types of information that have been the focus of substantial medical study, but have only recently started to play a bigger role in drug development, namely the metaedges Disease–associates–Gene, Compound–causes–Side Effect, Disease–presents–Symptom, Disease–localizes–Anatomy, and Gene–interacts–Gene.
The lower-middle tier contains the transcriptomics metaedges such as Compound–downregulates–Gene, Anatomy–expresses–Gene, Gene→regulates→Gene, and Disease–downregulates–Gene. Much excitement surrounds these resources due to their high throughput and genome-wide scope, which offers a route to drug discovery that is less biased by existing knowledge. However, our findings suggest that these resources are only moderately informative of drug efficacy. Other lower-middle tier metaedges were the product of time-intensive biological experimentation, such as Gene–participates–Pathway, Gene–participates–Molecular Function, and Gene–participates–Biological Process. Unlike the top tier resources, this knowledge has historically been pursued for basic science rather than primarily medical applications. The weak yet appreciable performance of the Gene–covaries–Gene suggests the synergy between the fields of evolutionary genomics and disease biology. The lower tier included the Gene–participates–Cellular Component metaedge, which may reflect that the relevance of cellular location to pharmacology is highly case dependent and not amenable to systematic profiling.
The performance of specific metapaths (Table 3) provides further insight. For example, significant emphasis has been put on the use of transcriptional data for drug repurposing . One common approach has been to identify compounds with opposing transcriptional signatures to a disease [18,70]. However, several systematic studies report underwhelming performance of this approach [24–26] — a finding supported by the low performance of the CuGdD and CdGuD metapaths in Project Rephetio. Nonetheless, other transcription-based methods showed some promise. Compounds with similar transcriptional signatures were prone to treating the same disease (CuGuCtD and CdGdCtD metapaths), while compounds with opposing transcriptional signatures were slightly averse to treating the same disease (CuGdCtD and CdGuCtD metapaths). In contrast, diseases with similar transcriptional profiles were not prone to treatment by the same compound (CtDdGuD and CtDuGdD).
By comparably assessing the informativeness of different metaedges and metapaths, Project Rephetio aims to guide future research towards promising data types and analyses. One caveat is that omics-scale experimental data will likely play a larger role in developing the next generation of pharmacotherapies. Hence, were performance reevaluated on treatments discovered in the forthcoming decades, the predictive ability of these data types may rise. Encouragingly, most data types were at least weakly informative and hence suitable for further study. Ideally, different data types would provide orthogonal information. However, our model for whether a compound treats a disease focused on 11 metapaths — a small portion of the hundreds of metapaths available. While parsimony aids interpretation, our model did not draw on the weakly-predictive high-throughput data types — which are intriguing for their novelty, scalability, and cost-effectiveness — as much as we had hypothesized.
Instead our model selected types of information traditionally considered in pharmacology. However unlike a pharmacologist whose area of expertise may be limited to a few drug classes, our model was able to predict probabilities of treatment for all 209,168 compound–disease pairs. Furthermore, our model systematically learned the importance of each type of network connectivity. For any compound–disease pair, we now can immediately provide the top network paths supporting its therapeutic efficacy. A traditional pharmacologist may be able to produce a similar explanation, but likely not until spending substantial time researching the compound’s pharmacology, the disease’s pathophysiology, and the molecular relationships in between. Accordingly, we hope certain predictions will spur further research, such as trials to investigate the off-label use of acamprosate for epilepsy, which is supported by one animal model .
As demonstrated by the 15 ictogenic compounds in our top 100 epilepsy predictions, Project Rephetio’s predictions can include contraindications in addition to indications. Since many of Hetionet v1.0’s relationship types are general (e.g. the Compound–binds–Gene relationship type conflates antagonist with agonist effects), we expect some high scoring predictions to exacerbate rather than treat the disease. However, the predictions made by Hetionet v1.0 represent such substantial relative enrichment over the null that uncovering the correct directionality is a logical next step and worth undertaking. Going forward, advances in automated mining of the scientific literature could enable extraction of precise relationship types at omics scale [72,73].
Future research should focus on gleaning orthogonal information from data types that are so expansive that computational methods are the only option. Our CuGuCtD feature — measuring whether a compound upregulates the same genes as compounds which treat the disease — is a good example. This metapath was informative by itself (Δ AUROC = 4.4%) but was not selected by the model, despite its orthogonal origin (gene expression) to selected metapaths. Using a more extensive catalog of treatments as the gold standard would be one possible approach to increase the power of feature selection.
Integrating more types of information into Hetionet should also be a future priority. The “network effect” phenomenon suggests that the addition of each new piece of information will enhance the value of Hetionet’s existing information. We envision a future where all biological knowledge is encoded into a single hetnet. Hetionet v1.0 was an early attempt, and we hope the strong performance of Project Rephetio in repurposing drugs foreshadows the many applications that will thrive from encoding biology in hetnets.
Hetionet was built entirely from publicly available resources with the goal of integrating a broad diversity of information types of medical relevance, ranging in scale from molecular to organismal. Practical considerations such as data availability, licensing, reusability, documentation, throughput, and standardization informed our choice of resources. We abided by a simple litmus test for determining how to encode information in a hetnet: nodes represent nouns, relationships represent verbs [74,75].
Our method for relationship prediction creates a strong incentive to avoid redundancy, which increases the computational burden without improving performance. In a previous study to predict disease–gene associations using a hetnet of pathophysiology , we found that different types of gene sets contributed highly redundant information. Therefore, in Hetionet v1.0 we reduced the number of gene set node types from 14 to 3 by omitting several gene set collections and aggregating all pathway nodes.
Nodes encode entities. We extracted nodes from standard terminologies, which provide curated vocabularies to enable data integration and prevent concept duplication. The ease of mapping external vocabularies, adoption, and comprehensiveness were primary selection criteria. Hetionet v1.0 includes nodes from 5 ontologies — which provide hierarchy of entities for a specific domain — selected for their conformity to current best practices .
We selected 137 terms from the Disease Ontology [77,78] (which we refer to as DO Slim [79,80]) as our disease set. Our goal was to identify complex diseases that are distinct and specific enough to be clinically relevant yet general enough to be well annotated. To this end, we included diseases that have been studied by GWAS and cancer types from
TopNodes_DOcancerslim . We ensured that no DO Slim disease was a subtype of another DO Slim disease. Symptoms were extracted from MeSH by taking the 438 descendants of Signs and Symptoms [82,83].
Approved small molecule compounds with documented chemical structures were extracted from DrugBank version 4.2 [84–86]. Unapproved compounds were excluded because our focus was repurposing. In addition, unapproved compounds tend to be less studied than approved compounds making them less attractive for our approach where robust network connectivity is critical. Finally, restricting to small molecules with known documented structures enabled us to map between compound vocabularies (see Mappings).
Side effects were extracted from SIDER version 4.1 [87–89]. SIDER codes side effects using UMLS identifiers , which we also adopted. Pharmacologic Classes were extracted from the DrugCentral data repository [91,92]. Only pharmacologic classes corresponding to the “Chemical/Ingredient”, “Mechanism of Action”, and “Physiologic Effect” FDA class types were included to avoid pharmacologic classes that were synonymous with indications .
Protein-coding human genes were extracted from Entrez Gene [93–95]. Anatomical structures, which we refer to as anatomies, were extracted from Uberon . We selected a subset of 402 Uberon terms by excluding terms known not to exist in humans and terms that were overly broad or arcane [97,98].
Pathways were extracted by combining human pathways from WikiPathways [99,100], Reactome , and the Pathway Interaction Database . The latter two resources were retrieved from Pathway Commons (RRID:SCR_002103) , which compiles pathways from several providers. Duplicate pathways and pathways without multiple participating genes were removed [104,105]. Biological processes, cellular components, and molecular functions were extracted from the Gene Ontology . Only terms with 2–1000 annotated genes were included.
Before adding relationships, all identifiers needed to be converted into the vocabularies matching that of our nodes. Oftentimes, our node vocabularies included external mappings. For example, the Disease Ontology includes mappings to MeSH, UMLS, and the ICD, several of which we submitted during the course of this study . In a few cases, the only option was to map using gene symbols, a disfavored method given that it can lead to ambiguities.
When mapping external disease concepts onto DO Slim, we used transitive closure. For example, the UMLS concept for primary progressive multiple sclerosis (
C0751964) was mapped to the DO Slim term for multiple sclerosis (
Chemical vocabularies presented the greatest mapping challenge , since these are poorly standardized . UniChem’s  Connectivity Search  was used to map compounds, which maps by atomic connectivity (based on First InChIKey Hash Blocks ) and ignores small molecular differences.
Anatomy–downregulates–Gene and Anatomy–upregulates–Gene edges [112–114] were extracted from Bgee , which computes differentially expressed genes by anatomy in post-juvenile adult humans. Anatomy–expresses–Gene edges were extracted from Bgee and TISSUES [116–118].
Compound–binds–Gene edges were aggregated from BindingDB [119,120], DrugBank [84,121], and DrugCentral . Only binding relationships to single proteins with affinities of at least 1 μM (as determined by Kd, Kᵢ, or IC₅₀) were selected from the October 2015 release of BindingDB [122,123]. Target, carrier, transporter, and enzyme interactions with single proteins (i.e. excluding protein groups) were extracted from DrugBank 4.2 [86,124]. In addition, all mapping DrugCentral target relationships were included .
Compound–treats–Disease (disease-modifying indications) and Compound–palliates–Disease (symptomatic indications) edges are from PharmacotherapyDB as described in Intermediate resources. Compound–causes–Side Effect edges were obtained from SIDER 4.1 [87–89], which uses natural language processing to identify side effects in drug labels. Compound–resembles–Compound relationships [86,125,126] represent chemical similarity and correspond to a Dice coefficient ≥ 0.5  between extended connectivity fingerprints [128,129]. Pharmacologic Class–includes–Compound edges were extracted from DrugCentral for three FDA class types [91,92]. Compound–downregulates–Gene and Compound–upregulates–Gene relationships were computed from LINCS L1000 as described in Intermediate resources.
Disease–associates–Gene edges were extracted from the GWAS Catalog , DISEASES [131,132], DisGeNET [133,134], and DOAF [135,136]. The GWAS Catalog compiles disease–SNP associations from published GWAS . We aggregated overlapping loci associated with each disease and identified the mode reported gene for each high confidence locus [138,139]. DISEASES integrates evidence of association from text mining, curated catalogs, and experimental data . Associations from DISEASES with integrated scores ≥ 2 were included after removing the contribution of DistiLD. DisGeNET integrates evidence from over 10 sources and reports a single score for each association [141,142]. Associations with scores ≥ 0.06 were included. DOAF mines Entrez Gene GeneRIFs (textual annotations of gene function) for disease mentions . Associations with 3 or more supporting GeneRIFs were included. Disease–downregulates–Gene and Disease–upregulates–Gene relationships [144,145] were computed using STARGEO as described in Intermediate resources.
Disease–localizes–Anatomy, Disease–presents–Symptom, and Disease–resembles–Disease edges were calculated from MEDLINE co-occurrence [82,146]. MEDLINE is a subset of 21 million PubMed articles for which designated human curators have assigned topics. When retrieving articles for a given topic (MeSH term), we activated two non-default search options as specified below:
majr for selecting only articles where the topic is major and
noexp for suppressing explosion (returning articles linked to MeSH subterms). We identified 4,161,769 articles with two or more disease topics; 696,252 articles with both a disease topic (
majr) and an anatomy topic (
noexp) ; and 363,928 articles with both a disease topic (
majr) and a symptom topic (
noexp). We used a Fisher’s exact test  to identify pairs of terms that occurred together more than would be expected by chance in their respective corpus. We included co-occurring terms with p < 0.005 in Hetionet v1.0.
Gene→regulates→Gene directed edges were generated from the LINCS L1000 genetic interference screens (see Intermediate resources) and indicate that knockdown or overexpression of the source gene significantly dysregulated the target gene [149,150]. Gene–covaries–Gene edges represent evolutionary rate covariation ≥ 0.75 [151–153]. Gene–interacts–Gene edges [154,155] represent when two genes produce physically-interacting proteins. We compiled these interactions from the Human Interactome Database [156–159], the Incomplete Interactome , and our previous study . Gene–participates–Biological Process, Gene–participates–Cellular Component, and Gene–participates–Molecular Function edges are from Gene Ontology annotations . As described in Intermediate resources, annotations were propagated [162,163]. Gene–participates–Pathway edges were included from the human pathway resources described in the Nodes section [104,105].
Whether a certain type of relationship has directionality is defined at the metaedge level. Directed metaedges are only necessary when they connect a metanode to itself and correspond to an asymmetric relationship. In the case of Hetionet v1.0, the sole directed metaedge was Gene→regulates→Gene. To demonstrate the implications of directionality, Hetionet v1.0 contains two relationships between the genes HADH and STAT1: HADH–interacts–STAT1 and HADH→regulates→STAT1. Both edges can be represented in the inverse orientation: STAT1–interacts–HADH and STAT1←regulates←HADH. However due to directed nature of the regulates relationship, STAT1→regulates→HADH is a distinct edge, which does not exist in the network. Similarly, HADH–associates–obesity and obesity–associates–HADH are inverse orientations of the same underlying undirected relationship. Accordingly, the following path exists in the network: obesity–associates–HADH→regulates→STAT1, which can also be inverted to STAT1←regulates←HADH–associates–obesity.
In the process of creating Hetionet, we produced several datasets with broad applicability that extended beyond Project Rephetio. These resources are referred to as intermediate resources and described below.
STARGEO is a nascent platform for annotating and meta-analyzing differential gene expression experiments . The STAR acronym stands for Search-Tag-Analyze Resources, while GEO refers to the Gene Expression Omnibus [165,166]. STARGEO is a layer on top of GEO that crowdsources sample annotation and automates meta-analysis.
Using STARGEO, we computed differentially expressed genes between healthy and diseased samples for 49 diseases [144,145]. First, we and others created case/control tags for 66 diseases. After combing through GEO series and tagging samples, 49 diseases had sufficient data for case-control meta-analysis: multiple series with at least 3 cases and 3 controls. For each disease, we performed a random effects meta-analysis on each gene to combine log₂ fold-change across series. These analyses incorporated 27,019 unique samples from 460 series on 107 platforms.
Differentially expressed genes (false discovery rate ≤ 0.05) were identified for each disease. The median number of upregulated genes per disease was 351 and the median number of downregulated genes was 340. Endogenous depression was the only of the 49 diseases without any significantly dysregulated genes.
LINCS L1000 profiled the transcriptional response to small molecule and genetic interference perturbations. To increase throughput, expression was only measured for 978 genes, which were selected for their ability to impute expression of the remaining genes. A single perturbation was often assayed under a variety of conditions including cell types, dosages, timepoints, and concentrations. Each condition generates a single signature of dysregulation z-scores. We further processed these signatures to fit into our approach [167,168].
First we computed consensus signatures — which meta-analyze multiple signatures to condense them into one — for DrugBank small molecules, Entrez genes, and all L1000 perturbations [149,150]. First, we discarded non-gold (non-replicating or indistinct) signatures. Then we meta-analyzed z-scores using Stouffer’s method. Each signature was weighted by its average Spearman’s correlation to other signatures, with a 0.05 minimum, to de-emphasize discordant signatures. Our signatures include the 978 measured genes and the 6,489 imputed genes from the “best inferred gene subset”. To identify significantly dysregulated genes, we selected genes using a Bonferroni cutoff of p = 0.05 and limited the number of imputed genes to 1,000.
The consensus signatures for genetic perturbations allowed us to assess various characteristics of the L1000 dataset. First, we looked at whether genetic interference dysregulated its target gene in the expected direction . Looking at measured z-scores for target genes, we found that the knockdown perturbations were highly reliable, while the overexpression perturbations were only moderately reliable with 36% of overexpression perturbations downregulating their target. However, imputed z-scores for target genes barely exceeded chance at responding in the expected direction to interference. Hence, we concluded that the imputation quality of LINCS L1000 is poor. However, when restricting to significantly dyseregulated targets, 22 out of 29 imputed genes responded in the expected direction. This provides some evidence that the directional fidelity of imputation is higher for significantly dysregulated genes. Finally, we found that the transcriptional signatures of knocking down and overexpressing the same gene were positively correlated 65% of the time, suggesting the presence of a general stress response .
Based on these findings, we performed additional filtering of signifcantly dysregulated genes when building Hetionet v1.0. Compound–down/up-regulates–Gene relationships were restricted to the 125 most significant per compound-direction-status combination (status refers to measured versus imputed). For genetic interference perturbations, we restricted to the 50 most significant genes per gene-direction-status combination and merged the remaining edges into a single Gene→regulates→Gene relationship type containing both knockdown and overexpression perturbations.
We created PharmacotherapyDB, an open catalog of drug therapies for disease [171–173]. Version 1.0 contains 755 disease-modifying therapies and 390 symptomatic therapies between 97 diseases and 601 compounds.
This resource was motivated by the need for a gold standard of medical indications to train and evaluate our approach. Initially, we identified four existing indication catalogs : MEDI-HPS which mined indications from RxNorm, SIDER 2, MedlinePlus, and Wikipedia ; LabeledIn which extracted indications from drug labels via human curation [176–178]; EHRLink which identified medication–problem pairs that clinicians linked together in electronic health records [179,180]; and indications from PREDICT, which were compiled from UMLS relationships, drugs.com, and drug labels . After mapping to DO Slim and DrugBank Slim, the four resources contained 1,388 distinct indications.
However, we noticed that many indications were palliative and hence problematic as a gold standard of pharmacotherapy for our in silico approach. Therefore, we recruited two practicing physicians to curate the 1,388 preliminary indications . After a pilot on 50 indications, we defined three classifications: disease modifying meaning a drug that therapeutically changes the underlying or downstream biology of the disease; symptomatic meaning a drug that treats a significant symptom of the disease; and non-indication meaning a drug that neither therapeutically changes the underlying or downstream biology nor treats a significant symptom of the disease. Both curators independently classified all 1,388 indications.
The two curators disagreed on 444 calls (Cohen’s κ = 49.9%). We then recruited a third practicing physician, who reviewed all 1,388 calls and created a detailed explanation of his methodology . We proceeded with the third curator’s calls as the consensus curation. The first two curators did have reservations with classifying steroids as disease modifying for autoimmune diseases. We ultimately considered that these indications met our definition of disease modifying, which is based on a pathophysiological rather than clinical standard. Accordingly, therapies we consider disease modifying may not be used to alter long-term disease course in the modern clinic due to a poor risk–benefit ratio.
We created a browser (http://git.dhimmel.com/gene-ontology/) to provide straightforward access to Gene Ontology annotations [162,163]. Our service provides annotations between Gene Ontology terms and Entrez Genes. The user chooses propagated/direct annotation and all/experimental evidence. Annotations are currently available for 37 species and downloadable as user-friendly TSV files.
We committed to openly releasing our data and analyses from the origin of the project . Our goals were to contribute to the advancement of science [183,184], maximize our impact [185,186], and enable reproducibility [187–189]. These objectives required publicly distributing and openly licensing Hetionet and Project Rephetio data and analyses [190,191].
Hetionet v1.0 integrates 29 resources (Table 4), but two resources were removed prior to the v1.0 release. Of the total 31 resources , five were United States government works not subject to copyright, and twelve had licenses that met the Open Definition of knowledge version 2.1. Four resources allowed only non-commercial reuse. Most problematic were the remaining nine resources that had no license — which equates to all rights reserved by default and forbids reuse  — and one resource that explicitly forbid redistribution.
|Entrez Gene||G||custom||1||RRID:SCR_002473 [93–95]|
|LabeledIn||CtD, CpD||custom||1||RRID:SCR_015667 [176–178]|
|MEDLINE||DlA, DpS, DrD||custom||1||RRID:SCR_002185 [82,146]|
|Pathway Interaction Database||PW, GpPW||1||RRID:SCR_006866 [102,104,105]|
|Disease Ontology||D||CC BY 3.0||2ᴼᴰ||RRID:SCR_000476 [77–80]|
|DISEASES||DaG||CC BY 4.0||2ᴼᴰ||RRID:SCR_015664 [131,132,140]|
|DrugCentral||PC, CbG, PCiC||CC BY 4.0||2ᴼᴰ||RRID:SCR_015663 [91,92]|
|Gene Ontology||BP, CC, MF, GpBP, GpCC, GpMF||CC BY 4.0||2ᴼᴰ||RRID:SCR_002811 [106,161–163]|
|GWAS Catalog||DaG||custom||2ᴼᴰ||RRID:SCR_012745 [130,137–139]|
|Reactome||PW, GpPW||custom||2ᴼᴰ||RRID:SCR_003485 [101,103–105]|
|LINCS L1000||CdG, CuG, Gr>G||custom||2ᴼᴰ||[149,150,195]|
|TISSUES||AeG||CC BY 4.0||2ᴼᴰ||RRID:SCR_015665 [116–118]|
|Uberon||A||CC BY 3.0||2ᴼᴰ||RRID:SCR_010668 [96–98]|
|WikiPathways||PW, GpPW||CC BY 3.0 / custom||2ᴼᴰ||RRID:SCR_002134 [99,100,104,105]|
|BindingDB||CbG||mixed CC BY 3.0 & CC BY-SA 3.0||2ᴼᴰ||RRID:SCR_000390 [119,120,122,123]|
|DrugBank||C, CbG, CrC||custom||2||RRID:SCR_002700 [84–86,196]|
|MEDI||CtD, CpD||CC BY-NC-SA 3.0||2||RRID:SCR_015668 [174,175]|
|PREDICT||CtD, CpD||CC BY-NC-SA 3.0||2||[24,174]|
|SIDER||SE, CcSE||CC BY-NC-SA 4.0||2||RRID:SCR_004321 [87–89]|
|Bgee||AeG, AdG, AuG||4||RRID:SCR_002028 [112–115]|
|Evolutionary Rate Covariation||GcG||4||RRID:SCR_015669 [151–153]|
|Human Interactome Database||GiG||4||RRID:SCR_015670 [154–159]|
Additional difficulty resulted from license incompatibles across resources, which was caused primarily by non-commercial and share-alike stipulations. Furthermore, it was often unclear who owned the data . Therefore, we sought input from legal experts and chronicled our progress [193,195–197,199].
Ultimately, we did not find an ideal solution. We had to choose between absolute compliance and Hetionet: strictly adhering to copyright and licensing arrangements would have decimated the network. On the other hand, in the United States, mere facts are not subject to copyright, and fair use doctrine helps protect reuse that is transformative and educational. Hence, we choose a path forward which balanced legal, normative, ethical, and scientific considerations.
If a resource was in the public domain, we licensed any derivatives as CC0 1.0. For resources licensed to allow reuse, redistribution, and modification, we transmitted their licenses as properties on the specific nodes and relationships in Hetionet v1.0. For all other resources — for example, resources without licenses or with licenses that forbid redistribution — we sent permission requests to their creators. The median time till first response to our permission requests was 16 days, with only 2 resources affirmatively granting us permission. We did not receive any responses asking us to remove a resource. However, we did voluntarily remove MSigDB , since its license was highly problematic . As a result of our experience, we recommend that publicly-funded data should be explicitly dedicated to the public domain whenever possible.
From Hetionet, we derived five permuted hetnets . The permutations preserve node degree but eliminate edge specificity by employing an algorithm called XSwap to randomly swap edges . To extend XSwap to hetnets , we permuted each metaedge separately, so that edges were only swapped with other edges of the same type. We adopted a Markov chain approach, whereby the first permuted hetnet was generated from Hetionet v1.0, the second permuted hetnet was generated from the first, and so on. For each metaedge, we assessed the percent of edges unchanged as the algorithm progressed to ensure that a sufficient number of swaps had been performed to randomize the network . Permuted hetnets are useful for computing the baseline performance of meaningless edges while preserving node degree . Since, our use of permutation focused on assessing Δ AUROC, a small number of permuted hetnets was sufficient, as the variability in a metapath’s AUROC across the permuted hetnets was low.
Traditional relational databases — such as SQLite, MySQL, and PostgreSQL — excel at storing highly structured data in tables. Connectivity between tables is accomplished using foreign-key references between columns. However, for many biomedical applications the connectivity between entities is of foremost importance. Furthermore, enforcing a rigid structure of what attributes an entity may possess is less important and often unnecessarily prohibitive. Graph databases focus instead on capturing connectivity (relationships) between entities (nodes). Accordingly, graph databases such as Neo4j offer greater ease when modeling biomedical relationships and superior performance when traversing many levels of connectivity [204,205]. Until recently, graph database adoption in bioinformatics was limited . However lately, the demand to model and capture biological connectivity at scale has led to increasing adoption [207–210].
We used the Neo4j graph database for storing and operating on Hetionet and noticed major benefits from tapping into this large open source ecosystem . Persistent storage with immediate access and the Cypher query language — a sort of SQL for hetnets — were two of the biggest benefits. To facilitate our migration to Neo4j, we updated
hetio — our existing Python package for hetnets  — to export networks into Neo4j and DWPC queries to Cypher. In addition, we created an interactive GraphGist for Project Rephetio, which introduces our approach and showcases its Cypher queries. Finally, we created a public Neo4j instance , which leverages several modern technologies such Neo4j Browser guides, cloud hosting with HTTPS, and Docker deployment [214,215].
Project Rephetio relied on the previously-published DWPC metric to generate features for compound–disease pairs. The DWPC measures the prevalence of a given metapath between a given source and target node . It is calculated by first extracting all paths from the source to target node that follow the specified metapath. Next, each path is weighted by taking the product of the node degrees along the path raised to a negative exponent. This damping exponent — the sole parameter — thereby determines the extent that paths through high-degree nodes are downweighted: we chose w = 0.4 based on our past optimizations . The DWPC equals the sum of the path weights (referred to as path-degree products). Traversing the hetnet to extract all paths between a source and target node, which we performed in Neo4j, is the most computationally intensive step in computing DWPCs . For future work, we are exploring matrix multiplication approaches, which could improve runtime several orders of magnitude.
Project Rephetio made several refinements to metapath-based hetnet edge prediction compared to previous studies [22,23]. First, we transformed DWPCs by mean scaling and then taking the inverse hyperbolic sine  to make them more amenable to modeling . Second, we bifurcated the workflow into an all-features stage and an all-observations stage . The all-features stage assesses feature performance and does not require computing features for all negatives. Here we selected a random subset of 3,020 (4 × 755) negatives. Little error was introduced by this optimization, since the predominant limitation to performance assessment was the small number of positives (755) rather than negatives. Based on the all-features performance assessment , we selected 142 DWPCs to compute on all observations (all 209,168 compound–disease pairs). The feature selection was designed to remove uninformative features (according to permutation) and guard against edge-dropout contamination . Third, we included 14 degree features, which assess the degree of a specific metaedge for either the source compound or target disease.
To improve the interpretability of the predictions, we developed a method for decomposing a prediction into its network support . This information is deployed to our Neo4j Browser guides, allowing users to assess the biomedical evidence contributing to a given prediction. First, we used logistic regression terms to quantify the contribution of metapaths that positively support a prediction. Second, we decomposed a metapath’s contribution, according to its DWPC, into specific paths contributions. Finally, we aggregated paths based on their source (first) or target (last) edge to quantify the contribution of specific edges of the source compound or target disease .
Using the acamprosate–epilepsy prediction as an example, we first quantified metapath contributions: 40% of the prediction was supported by CbGbCtD paths, 36% by CbGaD paths, 11% by CcSEcCtD paths, 8% by CbGpPWpGaD paths, and 5% by CbGeAlD paths. Second, we calculated path contributions: Acamprosate–binds–GRM5–associates–epilepsy syndrome was the most supportive path, contributing 11% of the prediction. Finally, we aggregated path contributions to calculate that the source edge of Acamprosate—binds—GRM5 contributed 23% of the prediction, while the target edge of epilepsy syndrome–treats–Felbamate contributed 12%.
The 755 treatments in Hetionet v1.0 are not evenly distributed between all compounds and diseases. For example, methotrexate treats 19 diseases and hypertension is treated by 68 compounds. We estimated a prior probability of treatment — based only on the treatment degree of the source compound and target disease — on 744,975 permutations of the bipartite treatment network . Methotrexate received a 79.6% prior probability of treating hypertension, whereas a compound and disease that both had only one treatment received a prior of 0.12%.
Across the 209,168 compound–disease pairs, the prior predicted the known treatments with AUROC = 97.9%. The strength of this association threatened to dominate our predictions. However, not modeling the prior can lead to omitted-variable bias and confounded proxy variables. To address the issue, we included the logit-transformed prior, without any regularization, as a term in the model. This restricted model fitting to the 29,799 observations with a nonzero prior — corresponding to the 387 compounds and 77 diseases with at least one treatment. To enable predictions for all 209,168 observations, we set the prior for each compound–disease pair to the overall prevalence of positives (0.36%).
This method succeeded at accommodating the treatment degrees. The prior probabilities performed poorly on the validation sets with AUROC = 54.1% on DrugCentral indications and AUROC = 62.5% on clinical trials. This performance dropoff compared to training shows the danger of encoding treatment degree into predictions. The benefits of our solution are highlighted by the superior validation performance of our predictions compared to the prior (Figure 3).
We evaluated our predictions on four sets of indications as shown in Figure 3.
Only the Clinical Trial and DrugCentral indication sets were used for external validation, since the Disease Modifying and Symptomatic indications were included in the hetnet. As an aside, several additional indication catalogs have recently been published, which future studies may want to also consider [174,226–228].
We conducted our study using Thinklab — a platform for realtime open collaborative science — on which this study was the first project . We began the study by publicly proposing the idea and inviting discussion . We continued by chronicling our progress via discussions. We used Thinklab as the frontend to coordinate and report our analyses and GitHub as the backend to host our code, data, and notebooks. On top of our Thinklab team consisting of core contributors, we welcomed community contribution and review. In areas where our expertise was lacking or advice would be helpful, we sought input from domain experts and encouraged them to respond on Thinklab where their comments would be CC BY licensed and their contribution rated and rewarded.
In total, 40 non-team members commented across 86 discussions, which generated 622 comments and 191 notes (Figure 6). Thinklab content for this project totaled 145,771 words or 918,837 characters . Using an estimated 7,000 words per academic publication as a benchmark, Project Rephetio generated written content comparable in volume to 20.8 publications prior to its completion. We noticed several other benefits from using Thinklab including forging a community of contributors ; receiving feedback during the early stages when feedback was most actionable ; disseminating our research without delay [233,234]; opening avenues for external input ; facilitating problem-oriented teaching [236,237]; and improving our documentation by maintaining a publication-grade digital lab notebook .
Thinklab began winding down operations in July 2017 and has switched to a static state. While users will no longer be able to add comments, the corpus of content remains browsable at https://think-lab.github.io and available in machine-readable formats at
The preprint for this study is available at doi.org/bs4f . The manuscript was written in markdown, originally on Thinklab at doi.org/bszr . In August 2017, we switched to using the Manubot system to generate the manuscript. With Manubot, a GitHub repository (
dhimmel/rephetio-manuscript) tracks the manuscript’s source code, while continuous integration automatically rebuilds the manuscript upon changes. As a result, the latest version of the manuscript is always available at dhimmel.github.io/rephetio-manuscript. Additionally, readers can leave feedback or questions for the Project Rephetio team via GitHub Issues.
All software and datasets from Project Rephetio are publicly available on GitHub, Zenodo, or Figshare . Additional documentation for these materials is available in the corresponding Thinklab discussions. For reader convenience, software, datasets, and Thinklab discussions have been cited throughout the manuscript as relevant.
We are immensely grateful to our Thinklab contributors who joined us in our experiment of radically open science. The following non-team members provided contributions that received 5 or more Thinklab points: Lars Juhl Jensen, Frederic Bastian, Alexander Pico, Casey Greene, Benjamin Good, Craig Knox, Mike Gilson, Chris Mungall, Katie Fortney, Venkat Malladi, Tudor Oprea, MacKenzie Smith, Caty Chung, Allison McCoy, Alexey Strokach, Ritu Khare, Greg Way, Marina Sirota, Raghavendran Partha, Oleg Ursu, Jesse Spaulding, Gaya Nadarajan, Alex Ratner, Scooter Morris, Alessandro Didonna, Alex Pankov, Tong Shu Li, and Janet Piñero. Additionally, the founder of Thinklab, Jesse Spaulding, supported community contributions and developed the platform with Project Rephetio’s needs in mind. We also appreciate DigitalOcean’s sponsorship the Hetionet Browser to cover its hosting costs. Finally, we would like to thank Neo Technology, whose staff provided excellent technical support.
This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant Number 1144247 to DSH. SEB is supported by NINDS/NIH Grant Number 5R01NS088155 and the Heidrich Family and Friends Foundation. DH is supported by the the National Cancer Institute of the National Institutes of Health under Award Number UH2CA203792 and the National Library of Medicine under Award Number 1U01LM012675. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
1. Innovation in the pharmaceutical industry: New estimates of R&D costs
Joseph A. DiMasi, Henry G. Grabowski, Ronald W. Hansen
Journal of Health Economics (2016-05) https://doi.org/10.1016/j.jhealeco.2016.01.012
2. A guide to drug discovery: Trends in development and approval times for new therapeutics in the United States
Janice M. Reichert
Nature Reviews Drug Discovery (2003-09) https://doi.org/10.1038/nrd1178
3. Clinical development success rates for investigational drugs
Michael Hay, David W Thomas, John L Craighead, Celia Economides, Jesse Rosenthal
Nature Biotechnology (2014-01-09) https://doi.org/10.1038/nbt.2786
4. Diagnosing the decline in pharmaceutical R&D efficiency
Jack W. Scannell, Alex Blanckley, Helen Boldon, Brian Warrington
Nature Reviews Drug Discovery (2012-03-01) https://doi.org/10.1038/nrd3681
5. Drug repositioning: identifying and developing new uses for existing drugs
Ted T. Ashburn, Karl B. Thor
Nature Reviews Drug Discovery (2004-08) https://doi.org/10.1038/nrd1468
6. A method for systematic discovery of adverse drug events from clinical notes
Guan Wang, Kenneth Jung, Rainer Winnenburg, Nigam H Shah
Journal of the American Medical Informatics Association (2015-07-31) https://doi.org/10.1093/jamia/ocv102
7. Validating drug repurposing signals using electronic health records: a case study of metformin associated with reduced cancer mortality
H. Xu, M. C. Aldrich, Q. Chen, H. Liu, N. B. Peterson, Q. Dai, M. Levy, A. Shah, X. Han, X. Ruan, … J. C. Denny
Journal of the American Medical Informatics Association (2014-07-22) https://doi.org/10.1136/amiajnl-2014-002649
8. Mining Retrospective Data for Virtual Prospective Drug Repurposing: L-DOPA and Age-related Macular Degeneration
Murray H. Brilliant, Kamyar Vaziri, Thomas B. Connor Jr., Stephen G. Schwartz, Joseph J. Carroll, Catherine A. McCarty, Steven J. Schrodi, Scott J. Hebbring, Krishna S. Kishor, Harry W. Flynn Jr., … Brian S. McKay
The American Journal of Medicine (2016-03) https://doi.org/10.1016/j.amjmed.2015.10.015
9. Data-Driven Prediction of Drug Effects and Interactions
N. P. Tatonetti, P. P. Ye, R. Daneshjou, R. B. Altman
Science Translational Medicine (2012-03-14) https://doi.org/10.1126/scitranslmed.3003377
10. Bayesian statistical methods for genetic association studies
Matthew Stephens, David J. Balding
Nature Reviews Genetics (2009-10) https://doi.org/10.1038/nrg2615
11. The complex genetics of multiple sclerosis: pitfalls and prospects
Brain (2008-05-18) https://doi.org/10.1093/brain/awn081
12. Magic shotguns versus magic bullets: selectively non-selective drugs for mood disorders and schizophrenia
Bryan L. Roth, Douglas J. Sheffler, Wesley K. Kroeze
Nature Reviews Drug Discovery (2004-04) https://doi.org/10.1038/nrd1346
13. Network pharmacology: the next paradigm in drug discovery
Andrew L Hopkins
Nature Chemical Biology (2008-10-20) https://doi.org/10.1038/nchembio.118
14. Network pharmacology
Andrew L Hopkins
Nature Biotechnology (2007-10) https://doi.org/10.1038/nbt1007-1110
15. How were new medicines discovered?
David C. Swinney, Jason Anthony
Nature Reviews Drug Discovery (2011-06-24) https://doi.org/10.1038/nrd3480
16. Drug discovery in the age of systems biology: the rise of computational approaches for data integration
Murat Iskar, Georg Zeller, Xing-Ming Zhao, Vera van Noort, Peer Bork
Current Opinion in Biotechnology (2012-08) https://doi.org/10.1016/j.copbio.2011.11.010
17. The Connectivity Map: a new tool for biomedical research
Nature Reviews Cancer (2007-01) https://doi.org/10.1038/nrc2044
18. Applications of Connectivity Map in drug discovery and development
Xiaoyan A. Qu, Deepak K. Rajpal
Drug Discovery Today (2012-12) https://doi.org/10.1016/j.drudis.2012.07.017
19. In silicomethods for drug repurposing and pharmacology
Rachel A. Hodos, Brian A. Kidd, Khader Shameer, Ben P. Readhead, Joel T. Dudley
Wiley Interdisciplinary Reviews: Systems Biology and Medicine (2016-04-15) https://doi.org/10.1002/wsbm.1337
20. Computational Drug Repositioning: From Data to Therapeutics
MR Hurle, L Yang, Q Xie, DK Rajpal, P Sanseau, P Agarwal
Clinical Pharmacology & Therapeutics (2013-01-15) https://doi.org/10.1038/clpt.2013.1
21. In silico drug repositioning – what we need to know
Zhichao Liu, Hong Fang, Kelly Reagan, Xiaowei Xu, Donna L. Mendrick, William Slikker Jr, Weida Tong
Drug Discovery Today (2013-02) https://doi.org/10.1016/j.drudis.2012.08.005
22. Heterogeneous Network Edge Prediction: A Data Integration Approach to Prioritize Disease-Associated Genes
Daniel S. Himmelstein, Sergio E. Baranzini
PLOS Computational Biology (2015-07-09) https://doi.org/10.1371/journal.pcbi.1004259
23. Co-author Relationship Prediction in Heterogeneous Bibliographic Networks
Yizhou Sun, Rick Barber, Manish Gupta, Charu C. Aggarwal, Jiawei Han
2011 International Conference on Advances in Social Networks Analysis and Mining (2011-07) https://doi.org/10.1109/asonam.2011.112
24. PREDICT: a method for inferring novel drug indications with application to personalized medicine
A. Gottlieb, G. Y. Stein, E. Ruppin, R. Sharan
Molecular Systems Biology (2014-04-16) https://doi.org/10.1038/msb.2011.26
25. Systematic evaluation of connectivity map for disease indications
Jie Cheng, Lun Yang, Vinod Kumar, Pankaj Agarwal
Genome Medicine (2014-12) https://doi.org/10.1186/s13073-014-0095-1
26. Network-based in silico drug efficacy screening
Emre Guney, Jörg Menche, Marc Vidal, Albert-László Barábasi
Nature Communications (2016-02-01) https://doi.org/10.1038/ncomms10331
27. A new method for computational drug repositioning using drug pairwise similarity
Jiao Li, Zhiyong Lu
2012 IEEE International Conference on Bioinformatics and Biomedicine (2012-10) https://doi.org/10.1109/bibm.2012.6392722
28. Systematic Evaluation of Drug–Disease Relationships to Identify Leads for Novel Drug Uses
AP Chiang, AJ Butte
Clinical Pharmacology & Therapeutics (2009-07-01) https://doi.org/10.1038/clpt.2009.103
29. The Connectivity Map: Using Gene-Expression Signatures to Connect Small Molecules, Genes, and Disease
Science (2006-09-29) https://doi.org/10.1126/science.1132939
30. Transcriptional data: a new gateway to drug repositioning?
Francesco Iorio, Timothy Rittman, Hong Ge, Michael Menden, Julio Saez-Rodriguez
Drug Discovery Today (2013-04) https://doi.org/10.1016/j.drudis.2012.07.014
31. The support of human genetic evidence for approved drug indications
Matthew R Nelson, Hannah Tipney, Jeffery L Painter, Judong Shen, Paola Nicoletti, Yufeng Shen, Aris Floratos, Pak Chung Sham, Mulin Jun Li, Junwen Wang, … Philippe Sanseau
Nature Genetics (2015-06-29) https://doi.org/10.1038/ng.3314
32. Use of genome-wide association studies for drug repositioning
Philippe Sanseau, Pankaj Agarwal, Michael R Barnes, Tomi Pastinen, J Brent Richards, Lon R Cardon, Vincent Mooser
Nature Biotechnology (2012-04-10) https://doi.org/10.1038/nbt.2151
33. Drug Target Identification Using Side-Effect Similarity
M. Campillos, M. Kuhn, A.-C. Gavin, L. J. Jensen, P. Bork
Science (2008-07-11) https://doi.org/10.1126/science.1158140
34. Computational drug repositioning based on side-effects mined from social media
Timothy Nugent, Vassilis Plachouras, Jochen L. Leidner
PeerJ Computer Science (2016-02-24) https://doi.org/10.7717/peerj-cs.46
35. Human symptoms–disease network
XueZhong Zhou, Jörg Menche, Albert-László Barabási, Amitabh Sharma
Nature Communications (2014-06-26) https://doi.org/10.1038/ncomms5212
36. Pathway-based Bayesian inference of drug–disease interactions
Naruemon Pratanwanich, Pietro Lió
Mol. BioSyst. (2014) https://doi.org/10.1039/c4mb00014e
37. Exploring the power of Hetionet: a Cypher query depot
ThinkLab (2016-06-25) https://doi.org/10.15363/thinklab.d220
38. Dhimmel/Hetionet V1.0.0: Hetionet V1.0 In Json, Tsv, And Neo4J Formats
Zenodo (2017-02-03) https://doi.org/10.5281/zenodo.268568
39. Computing standardized logistic regression coefficients
Daniel Himmelstein, Antoine Lizee
ThinkLab (2016-04-21) https://doi.org/10.15363/thinklab.d205
40. Our hetnet edge prediction methodology: the modeling framework for Project Rephetio
ThinkLab (2016-05-04) https://doi.org/10.15363/thinklab.d210
41. Dhimmel/Learn V1.0: The Machine Learning Repository For Project Rephetio
Zenodo (2017-02-04) https://doi.org/10.5281/zenodo.268654
42. Predictions of whether a compound treats a disease
Daniel Himmelstein, Chrissy Hessler, Pouya Khankhanian
ThinkLab (2016-05-17) https://doi.org/10.15363/thinklab.d203
43. Development of Novel Pharmacotherapeutics for Tobacco Dependence: Progress and Future Directions
D. Harmey, P. R. Griffin, P. J. Kenny
Nicotine & Tobacco Research (2012-09-27) https://doi.org/10.1093/ntr/nts201
44. Varenicline Is a Partial Agonist at 4beta2 and a Full Agonist at 7 Neuronal Nicotinic Receptors
K. B. Mihalak
Molecular Pharmacology (2006-06-20) https://doi.org/10.1124/mol.106.025130
45. A variant associated with nicotine dependence, lung cancer and peripheral arterial disease
Thorgeir E. Thorgeirsson, Frank Geller, Patrick Sulem, Thorunn Rafnar, Anna Wiste, Kristinn P. Magnusson, Andrei Manolescu, Gudmar Thorleifsson, Hreinn Stefansson, Andres Ingason, … Kari Stefansson
Nature (2008-04-03) https://doi.org/10.1038/nature06846
46. Evaluation of the safety of bupropion (Zyban) for smoking cessation from experience gained in general practice use in England in 2000
Andrew Boshier, Lynda V. Wilton, Saad A. W. Shakir
European Journal of Clinical Pharmacology (2003-12-01) https://doi.org/10.1007/s00228-003-0693-0
47. Efficacy and Safety of Varenicline for Smoking Cessation
J. Taylor Hays, Jon O. Ebbert, Amit Sood
The American Journal of Medicine (2008-04) https://doi.org/10.1016/j.amjmed.2008.01.017
48. Nicotine receptor partial agonists for smoking cessation
Kate Cahill, Nicola Lindson-Hawley, Kyla H Thomas, Thomas R Fanshawe, Tim Lancaster
Cochrane Database of Systematic Reviews (2016-05-09) https://doi.org/10.1002/14651858.cd006103.pub7
49. Placebo-Controlled Trial of Cytisine for Smoking Cessation
Robert West, Witold Zatonski, Magdalena Cedzynska, Dorota Lewandowska, Joanna Pazik, Paul Aveyard, John Stapleton
New England Journal of Medicine (2011-09-29) https://doi.org/10.1056/nejmoa1102035
50. Cytisine versus Nicotine for Smoking Cessation
Natalie Walker, Colin Howe, Marewa Glover, Hayden McRobbie, Joanne Barnes, Vili Nosa, Varsha Parag, Bruce Bassett, Christopher Bullen
New England Journal of Medicine (2014-12-18) https://doi.org/10.1056/nejmoa1407764
51. Repeated administration of an acetylcholinesterase inhibitor attenuates nicotine taking in rats and smoking behavior in human smokers
RL Ashare, BA Kimmey, LE Rupprecht, ME Bowers, MR Hayes, HD Schmidt
Translational Psychiatry (2016-01-19) https://doi.org/10.1038/tp.2015.209
52. Prediction in epilepsy
Pouya Khankhanian, Daniel Himmelstein
ThinkLab (2016-09-18) https://doi.org/10.15363/thinklab.d224
53. Visualizing the top epilepsy predictions in Cytoscape
Daniel Himmelstein, Pouya Khankhanian, Alexander Pico, Lars Juhl Jensen, Scooter Morris
ThinkLab (2017-01-24) https://doi.org/10.15363/thinklab.d230
54. Treatment of Refractory Status Epilepticus With Inhalational Anesthetic Agents Isoflurane and Desflurane
Seyed M. Mirsattari, Michael D. Sharpe, G. Bryan Young
Archives of Neurology (2004-08-01) https://doi.org/10.1001/archneur.61.8.1254
55. Anatomical Therapeutic Chemical Classification System (WHO)
The SAGE Encyclopedia of Pharmacology and Society (2016-03-29) https://doi.org/10.4135/9781483349985.n37
56. Antiepileptic Drug Interactions - Principles and Clinical Implications
Svein I. Johannessen, Cecilie Johannessen Landmark
Current Neuropharmacology (2010-09-01) https://doi.org/10.2174/157015910792246254
57. The neurobiology of antiepileptic drugs
Michael A. Rogawski, Wolfgang Löscher
Nature Reviews Neuroscience (2004-07) https://doi.org/10.1038/nrn1430
58. Proconvulsant effects of antidepressants — What is the current evidence?
Cecilie Johannessen Landmark, Oliver Henning, Svein I. Johannessen
Epilepsy & Behavior (2016-08) https://doi.org/10.1016/j.yebeh.2016.01.029
59. Why we predicted ictogenic tricyclic compounds treat epilepsy?
ThinkLab (2017-03-10) https://doi.org/10.15363/thinklab.d231
60. Antidepressants and seizures: Clinical anecdotes overshadow neuroscience
John W. Dailey, Dean K. Naritoku
Biochemical Pharmacology (1996-11) https://doi.org/10.1016/s0006-2952(96)00509-6
61. Movement disorders in patients taking anticonvulsants
C Zadikoff, RP Munhoz, AN Asante, N Politzer, R Wennberg, P Carlen, A Lang
Journal of Neurology, Neurosurgery & Psychiatry (2007-02-01) https://doi.org/10.1136/jnnp.2006.100222
62. Anticonvulsant-induced downbeat nystagmus in epilepsy
Dongyan Wu, Roland D. Thijs
Epilepsy & Behavior Case Reports (2015) https://doi.org/10.1016/j.ebcr.2015.07.003
63. The effect of antiepileptic drugs on visual performance
Emma J Roff Hilton, Sarah L Hosking, Tim Betts
Seizure (2003-05-30) https://doi.org/10.1016/s1059-1311(03)00082-7
64. Effect of antiepileptic drugs on sleep
Fabio Placidi, Anna Scalise, Maria Grazia Marciani, Andrea Romigi, Marina Diomedi, Gian Luigi Gigli
Clinical Neurophysiology (2000-09) https://doi.org/10.1016/s1388-2457(00)00411-9
65. Gastrointestinal adverse effects of antiepileptic drugs in intractable epileptic patients
Soodeh Razeghi Jahromi, Mansoureh Togha, Sohrab Hashemi Fesharaki, Masoumeh Najafi, Nahid Beladi Moghadam, Jalil Arab Kheradmand, Hadi Kazemi, Ali Gorji
Seizure (2011-05) https://doi.org/10.1016/j.seizure.2010.12.011
66. Methods for biological data integration: perspectives and challenges
Vladimir Gligorijević, Nataša Pržulj
Journal of The Royal Society Interface (2015-10-21) https://doi.org/10.1098/rsif.2015.0571
67. Multilayer networks
M. Kivela, A. Arenas, M. Barthelemy, J. P. Gleeson, Y. Moreno, M. A. Porter
Journal of Complex Networks (2014-07-14) https://doi.org/10.1093/comnet/cnu016
68. Renaming “heterogeneous networks” to a more concise and catchy term
Daniel Himmelstein, Casey Greene, Sergio Baranzini
ThinkLab (2015-08-16) https://doi.org/10.15363/thinklab.d104
69. Rephetio: Repurposing drugs on a hetnet [project]
Daniel Himmelstein, Antoine Lizee, Chrissy Hessler, Leo Brueggeman, Sabrina Chen, Dexter Hadley, Ari Green, Pouya Khankhanian, Sergio Baranzini
ThinkLab (2015-01-12) https://doi.org/10.15363/thinklab.4
70. Discovery and Preclinical Validation of Drug Indications Using Compendia of Public Gene Expression Data
M. Sirota, J. T. Dudley, J. Kim, A. P. Chiang, A. A. Morgan, A. Sweet-Cordero, J. Sage, A. J. Butte
Science Translational Medicine (2011-08-17) https://doi.org/10.1126/scitranslmed.3001318
71. Acamprosate attenuates the handling induced convulsions during alcohol withdrawal in Swiss Webster mice
Justin M. Farook, Ali Krazem, Ben Lewis, Dennis J. Morrell, John M. Littleton, Susan Barron
Physiology & Behavior (2008-09) https://doi.org/10.1016/j.physbeh.2008.05.020
72. Data programming with DDLite
Henry R. Ehrenberg, Jaeho Shin, Alexander J. Ratner, Jason A. Fries, Christopher Ré
Proceedings of the Workshop on Human-In-the-Loop Data Analytics - HILDA ’16 (2016) https://doi.org/10.1145/2939502.2939515
73. Brainstorming future directions for Hetionet
Daniel Himmelstein, Benjamin Good, Pouya Khankhanian, Alex Ratner
ThinkLab (2016-11-19) https://doi.org/10.15363/thinklab.d227
74. English, Chinese and ER diagrams
Peter Pin-Shan Chen
Data & Knowledge Engineering (1997-06) https://doi.org/10.1016/s0169-023x(97)00017-7
75. Data nomenclature: naming and abbreviating our network types
Daniel Himmelstein, Lars Juhl Jensen, Pouya Khankhanian
ThinkLab (2016-02-17) https://doi.org/10.15363/thinklab.d162
76. Ten Simple Rules for Selecting a Bio-ontology
James Malone, Robert Stevens, Simon Jupp, Tom Hancocks, Helen Parkinson, Cath Brooksbank
PLOS Computational Biology (2016-02-11) https://doi.org/10.1371/journal.pcbi.1004743
77. Disease Ontology: a backbone for disease semantic integration
L. M. Schriml, C. Arze, S. Nadendla, Y.-W. W. Chang, M. Mazaitis, V. Felix, G. Feng, W. A. Kibbe
Nucleic Acids Research (2011-11-12) https://doi.org/10.1093/nar/gkr972
78. Disease Ontology 2015 update: an expanded and updated database of human diseases for linking biomedical knowledge through disease data
W. A. Kibbe, C. Arze, V. Felix, E. Mitraka, E. Bolton, G. Fu, C. J. Mungall, J. X. Binder, J. Malone, D. Vasant, … L. M. Schriml
Nucleic Acids Research (2014-10-27) https://doi.org/10.1093/nar/gku1011
79. Unifying disease vocabularies
Daniel Himmelstein, Tong Shu Li
ThinkLab (2015-03-30) https://doi.org/10.15363/thinklab.d44
80. User-Friendly Extensions To The Disease Ontology V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/10.5281/zenodo.45584
81. Generating a focused view of disease ontology cancer terms for pan-cancer data integration and analysis
T.-J. Wu, L. M. Schriml, Q.-R. Chen, M. Colbert, D. J. Crichton, R. Finney, Y. Hu, W. A. Kibbe, H. Kincaid, D. Meerzaman, … R. Mazumder
Database (2015-04-04) https://doi.org/10.1093/database/bav032
82. Mining knowledge from MEDLINE articles and their indexed MeSH terms
Daniel Himmelstein, Alex Pankov
ThinkLab (2015-05-10) https://doi.org/10.15363/thinklab.d67
83. User-Friendly Extensions To Mesh V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/10.5281/zenodo.45586
84. DrugBank 4.0: shedding new light on drug metabolism
Vivian Law, Craig Knox, Yannick Djoumbou, Tim Jewison, An Chi Guo, Yifeng Liu, Adam Maciejewski, David Arndt, Michael Wilson, Vanessa Neveu, … David S. Wishart
Nucleic Acids Research (2013-11-06) https://doi.org/10.1093/nar/gkt1068
85. Unifying drug vocabularies
ThinkLab (2015-03-16) https://doi.org/10.15363/thinklab.d40
86. User-Friendly Extensions Of The Drugbank Database V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/10.5281/zenodo.45579
87. The SIDER database of drugs and side effects
Michael Kuhn, Ivica Letunic, Lars Juhl Jensen, Peer Bork
Nucleic Acids Research (2015-10-19) https://doi.org/10.1093/nar/gkv1075
88. Extracting side effects from SIDER 4
ThinkLab (2015-08-08) https://doi.org/10.15363/thinklab.d97
89. Extracting Tidy And User-Friendly Tsvs From Sider 4.1
Daniel S. Himmelstein
Zenodo (2016-02-03) https://doi.org/10.5281/zenodo.45521
90. The Unified Medical Language System (UMLS): integrating biomedical terminology
Nucleic Acids Research (2004-01-01) https://doi.org/10.1093/nar/gkh061
91. DrugCentral: online drug compendium
Oleg Ursu, Jayme Holmes, Jeffrey Knockel, Cristian G. Bologa, Jeremy J. Yang, Stephen L. Mathias, Stuart J. Nelson, Tudor I. Oprea
Nucleic Acids Research (2016-10-26) https://doi.org/10.1093/nar/gkw993
92. Incorporating DrugCentral data in our network
Daniel Himmelstein, Oleg Ursu, Mike Gilson, Pouya Khankhanian, Tudor Oprea
ThinkLab (2016-03-20) https://doi.org/10.15363/thinklab.d186
93. Entrez Gene: gene-centered information at NCBI
D. Maglott, J. Ostell, K. D. Pruitt, T. Tatusova
Nucleic Acids Research (2010-11-28) https://doi.org/10.1093/nar/gkq1237
94. Using Entrez Gene as our gene vocabulary
Daniel Himmelstein, Casey Greene, Alexander Pico
ThinkLab (2015-02-27) https://doi.org/10.15363/thinklab.d34
95. Processed Entrez Gene Datasets For Humans V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/10.5281/zenodo.45524
96. Uberon, an integrative multi-species anatomy ontology
Christopher J Mungall, Carlo Torniai, Georgios V Gkoutos, Suzanna E Lewis, Melissa A Haendel
Genome Biology (2012) https://doi.org/10.1186/gb-2012-13-1-r5
97. Tissue Node
Venkat Malladi, Daniel Himmelstein, Chris Mungall
ThinkLab (2015-03-19) https://doi.org/10.15363/thinklab.d41
98. User-Friendly Anatomical Structures Data From The Uberon Ontology V1.0
Daniel S. Himmelstein
Zenodo (2016-02-04) https://doi.org/10.5281/zenodo.45527
99. WikiPathways: capturing the full diversity of pathway knowledge
Martina Kutmon, Anders Riutta, Nuno Nunes, Kristina Hanspers, Egon L. Willighagen, Anwesha Bohler, Jonathan Mélius, Andra Waagmeester, Sravanthi R. Sinha, Ryan Miller, … Alexander R. Pico
Nucleic Acids Research (2015-10-19) https://doi.org/10.1093/nar/gkv1024
100. WikiPathways: Pathway Editing for the People
Alexander R Pico, Thomas Kelder, Martijn P van Iersel, Kristina Hanspers, Bruce R Conklin, Chris Evelo
PLoS Biology (2008-07-22) https://doi.org/10.1371/journal.pbio.0060184
101. The Reactome pathway Knowledgebase
Antonio Fabregat, Konstantinos Sidiropoulos, Phani Garapati, Marc Gillespie, Kerstin Hausmann, Robin Haw, Bijay Jassal, Steven Jupe, Florian Korninger, Sheldon McKay, … Peter D’Eustachio
Nucleic Acids Research (2015-12-09) https://doi.org/10.1093/nar/gkv1351
102. PID: the Pathway Interaction Database
Carl F. Schaefer, Kira Anthony, Shiva Krupa, Jeffrey Buchoff, Matthew Day, Timo Hannay, Kenneth H. Buetow
Nucleic Acids Research (2008-10-02) https://doi.org/10.1093/nar/gkn653
103. Pathway Commons, a web resource for biological pathway data
E. G. Cerami, B. E. Gross, E. Demir, I. Rodchenkov, O. Babur, N. Anwar, N. Schultz, G. D. Bader, C. Sander
Nucleic Acids Research (2010-11-10) https://doi.org/10.1093/nar/gkq1039
104. Adding pathway resources to your network
Alexander Pico, Daniel Himmelstein
ThinkLab (2015-06-08) https://doi.org/10.15363/thinklab.d72
105. Dhimmel/Pathways V2.0: Compiling Human Pathway Gene Sets
Daniel S. Himmelstein, Alexander R. Pico
Zenodo (2016-04-02) https://doi.org/10.5281/zenodo.48810
106. Gene Ontology: tool for the unification of biology
Michael Ashburner, Catherine A. Ball, Judith A. Blake, David Botstein, Heather Butler, J. Michael Cherry, Allan P. Davis, Kara Dolinski, Selina S. Dwight, Janan T. Eppig, … Gavin Sherlock
Nature Genetics (2000-05) https://doi.org/10.1038/75556
107. Disease Ontology feature requests
ThinkLab (2015-05-11) https://doi.org/10.15363/thinklab.d68
108. Chemical databases: curation or integration by user-defined equivalence?
Anne Hersey, Jon Chambers, Louisa Bellis, A. Patrícia Bento, Anna Gaulton, John P. Overington
Drug Discovery Today: Technologies (2015-07) https://doi.org/10.1016/j.ddtec.2015.01.005
109. UniChem: a unified chemical structure cross-referencing and identifier tracking system
Jon Chambers, Mark Davies, Anna Gaulton, Anne Hersey, Sameer Velankar, Robert Petryszak, Janna Hastings, Louisa Bellis, Shaun McGlinchey, John P Overington
Journal of Cheminformatics (2013) https://doi.org/10.1186/1758-2946-5-3
110. UniChem: extension of InChI-based compound mapping to salt, connectivity and stereochemistry layers
Jon Chambers, Mark Davies, Anna Gaulton, George Papadatos, Anne Hersey, John P Overington
Journal of Cheminformatics (2014-09-04) https://doi.org/10.1186/s13321-014-0043-5
111. InChI - the worldwide chemical structure identifier standard
Stephen Heller, Alan McNaught, Stephen Stein, Dmitrii Tchekhovskoi, Igor Pletnev
Journal of Cheminformatics (2013) https://doi.org/10.1186/1758-2946-5-7
112. Dhimmel/Bgee V1.0: Anatomy-Specific Gene Expression In Humans From Bgee
Daniel Himmelstein, Frederic Bastian, Sergio Baranzini
Zenodo (2016-03-08) https://doi.org/10.5281/zenodo.47157
113. Processing Bgee for tissue-specific gene presence and over/under-expression
Daniel Himmelstein, Frederic Bastian
ThinkLab (2015-11-03) https://doi.org/10.15363/thinklab.d124
114. Tissue-specific gene expression resources
Daniel Himmelstein, Frederic Bastian
ThinkLab (2015-06-17) https://doi.org/10.15363/thinklab.d81
115. Bgee: Integrating and Comparing Heterogeneous Transcriptome Data Among Species
Frederic Bastian, Gilles Parmentier, Julien Roux, Sebastien Moretti, Vincent Laudet, Marc Robinson-Rechavi
Lecture Notes in Computer Science (2008-06) https://doi.org/10.1007/978-3-540-69828-9_12
116. Comprehensive comparison of large-scale tissue expression datasets
Alberto Santos, Kalliopi Tsafou, Christian Stolte, Sune Pletscher-Frankild, Seán I. O’Donoghue, Lars Juhl Jensen
PeerJ (2015-06-30) https://doi.org/10.7717/peerj.1054
117. Gene–Tissue Relationships From The Tissues Database
Daniel Himmelstein, Lars Juhl Jensen
Zenodo (2015-08-09) https://doi.org/10.5281/zenodo.27244
118. The TISSUES resource for the tissue-specificity of genes
Daniel Himmelstein, Lars Juhl Jensen
ThinkLab (2015-07-10) https://doi.org/10.15363/thinklab.d91
119. BindingDB: A Web-Accessible Molecular Recognition Database
Xi Chen, Ming Liu, Michael Gilson
Combinatorial Chemistry & High Throughput Screening (2001-12-01) https://doi.org/10.2174/1386207013330670
120. BindingDB in 2015: A public database for medicinal chemistry, computational chemistry and systems pharmacology
Michael K. Gilson, Tiqing Liu, Michael Baitaluk, George Nicola, Linda Hwang, Jenny Chong
Nucleic Acids Research (2015-10-19) https://doi.org/10.1093/nar/gkv1072
121. DrugBank: a comprehensive resource for in silico drug discovery and exploration
D. S. Wishart
Nucleic Acids Research (2006-01-01) https://doi.org/10.1093/nar/gkj067
122. Integrating drug target information from BindingDB
Daniel Himmelstein, Mike Gilson
ThinkLab (2015-04-13) https://doi.org/10.15363/thinklab.d53
123. Processing The October 2015 Bindingdb
Daniel Himmelstein, Michael Gilson, Sergio Baranzini
Zenodo (2015-11-19) https://doi.org/10.5281/zenodo.33987
124. Protein (target, carrier, transporter, and enzyme) interactions in DrugBank
Daniel Himmelstein, Sabrina Chen
ThinkLab (2015-05-09) https://doi.org/10.15363/thinklab.d65
125. Calculating molecular similarities between DrugBank compounds
Daniel Himmelstein, Sabrina Chen
ThinkLab (2015-05-18) https://doi.org/10.15363/thinklab.d70
126. Pairwise molecular similarities between DrugBank compounds
Daniel Himmelstein, Leo Brueggeman, Sergio Baranzini
Figshare (2015) https://doi.org/10.6084/m9.figshare.1418386
127. Measures of the Amount of Ecologic Association Between Species
Lee R. Dice
Ecology (1945-07) https://doi.org/10.2307/1932409
128. Extended-Connectivity Fingerprints
David Rogers, Mathew Hahn
Journal of Chemical Information and Modeling (2010-05-24) https://doi.org/10.1021/ci100050t
129. The Generation of a Unique Machine Description for Chemical Structures-A Technique Developed at Chemical Abstracts Service.
H. L. Morgan
Journal of Chemical Documentation (1965-05) https://doi.org/10.1021/c160017a018
130. Dhimmel/Gwas-Catalog V1.0: Extracting Gene–Disease Associations From The Gwas Catalog
Daniel S. Himmelstein, Sergio E. Baranzini
Zenodo (2016-03-26) https://doi.org/10.5281/zenodo.48428
131. Processing the DISEASES resource for disease–gene relationships
Daniel Himmelstein, Lars Juhl Jensen
ThinkLab (2015-08-20) https://doi.org/10.15363/thinklab.d106
132. Dhimmel/Diseases V1.0: Processing The Diseases Database Of Gene–Disease Associations
Daniel S. Himmelstein, Lars Juhl Jensen
Zenodo (2016-03-26) https://doi.org/10.5281/zenodo.48425
133. Processing DisGeNET for disease-gene relationships
Daniel Himmelstein, janet piñero
ThinkLab (2015-08-17) https://doi.org/10.15363/thinklab.d105
134. Dhimmel/Disgenet V1.0: Processing The Disgenet Database Of Gene–Disease Associations
Daniel S. Himmelstein, Janet Piñero
Zenodo (2016-03-26) https://doi.org/10.5281/zenodo.48426
135. Functional disease annotations for genes using DOAF
ThinkLab (2015-07-14) https://doi.org/10.15363/thinklab.d94
136. Dhimmel/Doaf V1.0: Processing The Doaf Database Of Gene–Disease Associations
Daniel S. Himmelstein
Zenodo (2016-03-26) https://doi.org/10.5281/zenodo.48427
137. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog)
Jacqueline MacArthur, Emily Bowler, Maria Cerezo, Laurent Gil, Peggy Hall, Emma Hastings, Heather Junkins, Aoife McMahon, Annalisa Milano, Joannella Morales, … Helen Parkinson
Nucleic Acids Research (2016-11-29) https://doi.org/10.1093/nar/gkw1133
138. Extracting disease-gene associations from the GWAS Catalog
ThinkLab (2015-06-16) https://doi.org/10.15363/thinklab.d80
139. Calculating genomic windows for GWAS lead SNPs
Daniel Himmelstein, Marina Sirota, Greg Way
ThinkLab (2015-06-08) https://doi.org/10.15363/thinklab.d71
140. DISEASES: Text mining and data integration of disease–gene associations
Sune Pletscher-Frankild, Albert Pallejà, Kalliopi Tsafou, Janos X. Binder, Lars Juhl Jensen
Methods (2015-03) https://doi.org/10.1016/j.ymeth.2014.11.020
141. DisGeNET: a discovery platform for the dynamical exploration of human diseases and their genes
J. Pinero, N. Queralt-Rosinach, A. Bravo, J. Deu-Pons, A. Bauer-Mehren, M. Baron, F. Sanz, L. I. Furlong
Database (2015-04-15) https://doi.org/10.1093/database/bav028
142. DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants
Janet Piñero, Àlex Bravo, Núria Queralt-Rosinach, Alba Gutiérrez-Sacristán, Jordi Deu-Pons, Emilio Centeno, Javier García-García, Ferran Sanz, Laura I. Furlong
Nucleic Acids Research (2016-10-19) https://doi.org/10.1093/nar/gkw943
143. A Framework for Annotating Human Genome in Disease Context
Wei Xu, Huisong Wang, Wenqing Cheng, Dong Fu, Tian Xia, Warren A. Kibbe, Simon M. Lin
PLoS ONE (2012-12-10) https://doi.org/10.1371/journal.pone.0049686
144. STARGEO: expression signatures for disease using crowdsourced GEO annotation
Daniel Himmelstein, Frederic Bastian, Dexter Hadley, Casey Greene
ThinkLab (2015-07-28) https://doi.org/10.15363/thinklab.d96
145. Dhimmel/Stargeo V1.0: Differentially Expressed Genes For 48 Diseases From Stargeo
Daniel Himmelstein, Dexter Hadley, Alexander Schepanovski
Zenodo (2016-03-03) https://doi.org/10.5281/zenodo.46866
146. Dhimmel/Medline V1.0: Disease, Symptom, And Anatomy Cooccurence In Medline
Daniel S. Himmelstein
Zenodo (2016-03-28) https://doi.org/10.5281/zenodo.48445
147. Disease similarity from MEDLINE topic cooccurrence
ThinkLab (2015-07-14) https://doi.org/10.15363/thinklab.d93
148. On the Interpretation of χ 2 from Contingency Tables, and the Calculation of P
R. A. Fisher
Journal of the Royal Statistical Society (1922-01) https://doi.org/10.2307/2340521
149. Computing consensus transcriptional profiles for LINCS L1000 perturbations
Daniel Himmelstein, Caty Chung
ThinkLab (2015-03-26) https://doi.org/10.15363/thinklab.d43
150. Consensus signatures for LINCS L1000 perturbations
Daniel Himmelstein, Leo Brueggeman, Sergio Baranzini
Figshare (2016) https://doi.org/10.6084/m9.figshare.3085426.v1
151. Evolutionary Signatures amongst Disease Genes Permit Novel Methods for Gene Prioritization and Construction of Informative Gene-Based Networks
Nolan Priedigkeit, Nicholas Wolfe, Nathan L. Clark
PLOS Genetics (2015-02-13) https://doi.org/10.1371/journal.pgen.1004967
152. Selecting informative ERC (evolutionary rate covariation) values between genes
Daniel Himmelstein, Raghavendran Partha
ThinkLab (2015-04-22) https://doi.org/10.15363/thinklab.d57
153. Dhimmel/Erc V1.0: Processing Human Evolutionary Rate Covaration Data
Daniel S. Himmelstein
Zenodo (2016-03-28) https://doi.org/10.5281/zenodo.48444
154. Creating a catalog of protein interactions
Daniel Himmelstein, Dexter Hadley, Alexey Strokach
ThinkLab (2015-07-01) https://doi.org/10.15363/thinklab.d85
155. Dhimmel/Ppi V1.0: Compiling A Human Protein Interaction Catalog
Daniel S. Himmelstein, Sergio E. Baranzini
Zenodo (2016-03-28) https://doi.org/10.5281/zenodo.48443
156. Towards a proteome-scale map of the human protein–protein interaction network
Jean-François Rual, Kavitha Venkatesan, Tong Hao, Tomoko Hirozane-Kishikawa, Amélie Dricot, Ning Li, Gabriel F. Berriz, Francis D. Gibbons, Matija Dreze, Nono Ayivi-Guedehoussou, … Marc Vidal
Nature (2005-09-28) https://doi.org/10.1038/nature04209
157. An empirical framework for binary interactome mapping
Kavitha Venkatesan, Jean-François Rual, Alexei Vazquez, Ulrich Stelzl, Irma Lemmens, Tomoko Hirozane-Kishikawa, Tong Hao, Martina Zenkner, Xiaofeng Xin, Kwang-Il Goh, … Marc Vidal
Nature Methods (2008-12-07) https://doi.org/10.1038/nmeth.1280
158. Next-generation sequencing to generate interactome datasets
Haiyuan Yu, Leah Tardivo, Stanley Tam, Evan Weiner, Fana Gebreab, Changyu Fan, Nenad Svrzikapa, Tomoko Hirozane-Kishikawa, Edward Rietman, Xinping Yang, … Marc Vidal
Nature Methods (2011-04-24) https://doi.org/10.1038/nmeth.1597
159. A Proteome-Scale Map of the Human Interactome Network
Thomas Rolland, Murat Taşan, Benoit Charloteaux, Samuel J. Pevzner, Quan Zhong, Nidhi Sahni, Song Yi, Irma Lemmens, Celia Fontanillo, Roberto Mosca, … Marc Vidal
Cell (2014-11) https://doi.org/10.1016/j.cell.2014.10.050
160. Uncovering disease-disease relationships through the incomplete interactome
J. Menche, A. Sharma, M. Kitsak, S. D. Ghiassian, M. Vidal, J. Loscalzo, A.-L. Barabasi
Science (2015-02-19) https://doi.org/10.1126/science.1257601
161. The GOA database: Gene Ontology annotation updates for 2015
R. P. Huntley, T. Sawford, P. Mutowo-Meullenet, A. Shypitsyna, C. Bonilla, M. J. Martin, C. O’Donovan
Nucleic Acids Research (2014-11-06) https://doi.org/10.1093/nar/gku1113
162. Compiling Gene Ontology annotations into an easy-to-use format
Daniel Himmelstein, Casey Greene, Venkat Malladi, Frederic Bastian
ThinkLab (2015-03-12) https://doi.org/10.15363/thinklab.d39
163. Gene-Ontology: Initial Zenodo Release
Daniel Himmelstein, Casey Greene, Venkat Malladi, Frederic Bastian, Sergio Baranzini
Zenodo (2015-07-28) https://doi.org/10.5281/zenodo.21711
164. Precision annotation of digital samples in NCBI’s gene expression omnibus
Dexter Hadley, James Pan, Osama El-Sayed, Jihad Aljabban, Imad Aljabban, Tej D. Azad, Mohamad O. Hadied, Shuaib Raza, Benjamin Abhishek Rayikanti, Bin Chen, … Atul J. Butte
Scientific Data (2017-09-19) https://doi.org/10.1038/sdata.2017.125
165. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository
Nucleic Acids Research (2002-01-01) https://doi.org/10.1093/nar/30.1.207
166. NCBI GEO: archive for functional genomics data sets–update
T. Barrett, S. E. Wilhite, P. Ledoux, C. Evangelista, I. F. Kim, M. Tomashevsky, K. A. Marshall, K. H. Phillippy, P. M. Sherman, M. Holko, … A. Soboleva
Nucleic Acids Research (2012-11-27) https://doi.org/10.1093/nar/gks1193
167. Dhimmel/Lincs V2.0: Refined Consensus Signatures From Lincs L1000
Daniel Himmelstein, Leo Brueggeman, Sergio Baranzini
Zenodo (2016-03-08) https://doi.org/10.5281/zenodo.47223
168. l1000.db: SQLite database of LINCS L1000 metadata
Daniel Himmelstein, Leo Brueggeman, Sergio Baranzini
Figshare (2016) https://doi.org/10.6084/m9.figshare.3085837.v1
169. Assessing the imputation quality of gene expression in LINCS L1000
ThinkLab (2016-03-11) https://doi.org/10.15363/thinklab.d185
170. Positive correlations between knockdown and overexpression profiles from LINCS L1000
Daniel Himmelstein, Casey Greene, Lars Juhl Jensen
ThinkLab (2016-02-26) https://doi.org/10.15363/thinklab.d171
171. Announcing PharmacotherapyDB: the Open Catalog of Drug Therapies for Disease
ThinkLab (2016-03-15) https://doi.org/10.15363/thinklab.d182
172. PharmacotherapyDB 1.0: the open catalog of drug therapies for disease
Daniel Himmelstein, Pouya Khankhanian, Christine S. Hessler, Ari J. Green, Sergio Baranzini
Figshare (2016) https://doi.org/10.6084/m9.figshare.3103054
173. Dhimmel/Indications V1.0. Pharmacotherapydb: The Open Catalog Of Drug Therapies For Disease
Daniel S. Himmelstein, Pouya Khankhanian, Christine S. Hessler, Ari J. Green, Sergio E. Baranzini
Zenodo (2016-03-15) https://doi.org/10.5281/zenodo.47664
174. How should we construct a catalog of drug indications?
Daniel Himmelstein, Benjamin Good, Tudor Oprea, Allison McCoy, Antoine Lizee
ThinkLab (2015-01-13) https://doi.org/10.15363/thinklab.d21
175. Development and evaluation of an ensemble resource linking medications to their indications
Wei-Qi Wei, Robert M Cronin, Hua Xu, Thomas A Lasko, Lisa Bastarache, Joshua C Denny
Journal of the American Medical Informatics Association (2013-09) https://doi.org/10.1136/amiajnl-2012-001431
176. LabeledIn: Cataloging labeled indications for human drugs
Ritu Khare, Jiao Li, Zhiyong Lu
Journal of Biomedical Informatics (2014-12) https://doi.org/10.1016/j.jbi.2014.08.004
177. Scaling drug indication curation through crowdsourcing
R. Khare, J. D. Burger, J. S. Aberdeen, D. W. Tresner-Kirsch, T. J. Corrales, L. Hirchman, Z. Lu
Database (2015-03-22) https://doi.org/10.1093/database/bav016
178. Processing LabeledIn to extract indications
Daniel Himmelstein, Ritu Khare
ThinkLab (2015-04-02) https://doi.org/10.15363/thinklab.d46
179. Development and evaluation of a crowdsourcing methodology for knowledge base construction: identifying relationships between clinical problems and medications
Allison B McCoy, Adam Wright, Archana Laxmisan, Madelene J Ottosen, Jacob A McCoy, David Butten, Dean F Sittig
Journal of the American Medical Informatics Association (2012-09) https://doi.org/10.1136/amiajnl-2012-000852
180. Extracting indications from the ehrlink resource
ThinkLab (2015-05-01) https://doi.org/10.15363/thinklab.d62
181. Expert curation of our indication catalog for disease-modifying treatments
Daniel Himmelstein, Pouya Khankhanian, Chrissy Hessler
ThinkLab (2015-07-14) https://doi.org/10.15363/thinklab.d95
182. Enabling reproducibility and reuse
Jesse Spaulding, Daniel Himmelstein, Casey Greene, Benjamin Good
ThinkLab (2015-01-16) https://doi.org/10.15363/thinklab.d23
183. The need and drive for open data in biomedical publishing
Serials: The Journal for the Serials Community (2011-03-01) https://doi.org/10.1629/2431
184. The Open Knowledge Foundation: Open Data Means Better Science
Jennifer C. Molloy
PLoS Biology (2011-12-06) https://doi.org/10.1371/journal.pbio.1001195
185. How open science helps researchers succeed
Erin C McKiernan, Philip E Bourne, C Titus Brown, Stuart Buck, Amye Kenall, Jennifer Lin, Damon McDougall, Brian A Nosek, Karthik Ram, Courtney K Soderberg, … Tal Yarkoni
eLife (2016-07-07) https://doi.org/10.7554/elife.16800
186. Data reuse and the open data citation advantage
Heather A. Piwowar, Todd J. Vision
PeerJ (2013-10-01) https://doi.org/10.7717/peerj.175
187. Enhancing reproducibility for computational methods
V. Stodden, M. McNutt, D. H. Bailey, E. Deelman, Y. Gil, B. Hanson, M. A. Heroux, J. P. A. Ioannidis, M. Taufer
Science (2016-12-08) https://doi.org/10.1126/science.aah6168
188. Best Practices for Computational Science: Software Infrastructure and Environments for Reproducible and Extensible Research
Victoria Stodden, Sheila Miguez
Journal of Open Research Software (2014-07-09) https://doi.org/10.5334/jors.ay
189. Disclose all data in publications
Nature (2010-09-23) https://doi.org/10.1038/467401b
190. Open by default: a proposed copyright license and waiver agreement for open access research and data in peer-reviewed journals
Iain Hrynaszkiewicz, Matthew J Cockerill
BMC Research Notes (2012) https://doi.org/10.1186/1756-0500-5-494
191. Creative Commons licenses and the non-commercial condition: Implications for the re-use of biodiversity information
Gregor Hagedorn, Daniel Mietchen, Robert Morris, Donat Agosti, Lyubomir Penev, Walter Berendsohn, Donald Hobern
ZooKeys (2011-11-28) https://doi.org/10.3897/zookeys.150.2189
192. One network to rule them all
Daniel Himmelstein, Lars Juhl Jensen
ThinkLab (2015-08-14) https://doi.org/10.15363/thinklab.d102
193. Integrating resources with disparate licensing into an open network
Daniel Himmelstein, Lars Juhl Jensen, MacKenzie Smith, Katie Fortney, Caty Chung
ThinkLab (2015-08-28) https://doi.org/10.15363/thinklab.d107
194. Legal confusion threatens to slow data science
Nature (2016-08-03) https://doi.org/10.1038/536016a
195. LINCS L1000 licensing
ThinkLab (2015-09-28) https://doi.org/10.15363/thinklab.d110
Daniel Himmelstein, Katie Fortney, Craig Knox, Christopher Southan
ThinkLab (2016-05-08) https://doi.org/10.15363/thinklab.d213
197. Incomplete Interactome licensing
ThinkLab (2015-10-01) https://doi.org/10.15363/thinklab.d111
198. Who owns scientific data? The impact of intellectual property rights on the scientific publication chain
Learned Publishing (2005-04) https://doi.org/10.1087/0953151053584984
199. MSigDB licensing
ThinkLab (2015-09-28) https://doi.org/10.15363/thinklab.d108
200. Molecular signatures database (MSigDB) 3.0
A. Liberzon, A. Subramanian, R. Pinchback, H. Thorvaldsdottir, P. Tamayo, J. P. Mesirov
Bioinformatics (2011-05-05) https://doi.org/10.1093/bioinformatics/btr260
201. Assessing the effectiveness of our hetnet permutations
ThinkLab (2016-02-25) https://doi.org/10.15363/thinklab.d178
202. Randomization Techniques for Graphs
Sami Hanhijärvi, Gemma C. Garriga, Kai Puolamäki
Proceedings of the 2009 SIAM International Conference on Data Mining (2009-04-30) https://doi.org/10.1137/1.9781611972795.67
203. Permuting hetnets and implementing randomized edge swaps in cypher
ThinkLab (2015-12-21) https://doi.org/10.15363/thinklab.d136
204. Use of Graph Database for the Integration of Heterogeneous Biological Data
Byoung-Ha Yoon, Seon-Kyu Kim, Seon-Young Kim
Genomics & Informatics (2017) https://doi.org/10.5808/gi.2017.15.1.19
205. Comparative analysis of Relational and Graph databases
IOSR Journal of Engineering (2013-08) https://doi.org/10.9790/3021-03822527
206. Are graph databases ready for bioinformatics?
C. T. Have, L. J. Jensen
Bioinformatics (2013-10-17) https://doi.org/10.1093/bioinformatics/btt549
207. Representing and querying disease networks using graph databases
Artem Lysenko, Irina A. Roznovăţ, Mansoor Saqi, Alexander Mazein, Christopher J Rawlings, Charles Auffray
BioData Mining (2016-07-25) https://doi.org/10.1186/s13040-016-0102-8
208. Recon2Neo4j: applying graph database technologies for managing comprehensive genome-scale networks
Irina Balaur, Alexander Mazein, Mansoor Saqi, Artem Lysenko, Christopher J. Rawlings, Charles Auffray
Bioinformatics (2016-12-19) https://doi.org/10.1093/bioinformatics/btw731
209. The Network Library: a framework to rapidly integrate network biology resources
Georg Summer, Thomas Kelder, Marijana Radonjic, Marc van Bilsen, Suzan Wopereis, Stephane Heymans
Bioinformatics (2016-09-01) https://doi.org/10.1093/bioinformatics/btw436
210. The Monarch Initiative: an integrative data and analytic platform connecting phenotypes to genotypes across species
Christopher J. Mungall, Julie A. McMurry, Sebastian Köhler, James P. Balhoff, Charles Borromeo, Matthew Brush, Seth Carbon, Tom Conlin, Nathan Dunn, Mark Engelstad, … Melissa A. Haendel
Nucleic Acids Research (2016-11-29) https://doi.org/10.1093/nar/gkw1128
211. Using the neo4j graph database for hetnets
ThinkLab (2015-10-02) https://doi.org/10.15363/thinklab.d112
212. dhimmel/hetio v0.2.0: Neo4j export, Cypher query creation, hetnet stats, and other enhancements
Zenodo (2016-09-05) https://doi.org/10.5281/zenodo.61571
213. Hosting Hetionet in the cloud: creating a public Neo4j instance
ThinkLab (2016-06-23) https://doi.org/10.15363/thinklab.d216
214. Bioboxes: standardised containers for interchangeable bioinformatics software
Peter Belmann, Johannes Dröge, Andreas Bremges, Alice C. McHardy, Alexander Sczyrba, Michael D. Barton
GigaScience (2015-10-15) https://doi.org/10.1186/s13742-015-0087-0
215. Reproducibility of computational workflows is automated using continuous analysis
Brett K Beaulieu-Jones, Casey S Greene
Nature Biotechnology (2017-03-13) https://doi.org/10.1038/nbt.3780
216. Estimating the complexity of hetnet traversal
Daniel Himmelstein, Antoine Lizee
ThinkLab (2016-03-22) https://doi.org/10.15363/thinklab.d187
217. Alternative Transformations to Handle Extreme Values of the Dependent Variable
John B. Burbidge, Lonnie Magee, A. Leslie Robb
Journal of the American Statistical Association (1988-03) https://doi.org/10.2307/2288929
218. Transforming DWPCs for hetnet edge prediction
Daniel Himmelstein, Pouya Khankhanian, Antoine Lizee
ThinkLab (2016-04-01) https://doi.org/10.15363/thinklab.d193
219. Assessing the informativeness of features
ThinkLab (2015-10-04) https://doi.org/10.15363/thinklab.d115
220. Edge dropout contamination in hetnet edge prediction
ThinkLab (2016-05-16) https://doi.org/10.15363/thinklab.d215
221. Decomposing predictions into their network support
ThinkLab (2016-12-21) https://doi.org/10.15363/thinklab.d229
222. Decomposing the DWPC to assess intermediate node or edge contributions
ThinkLab (2016-12-15) https://doi.org/10.15363/thinklab.d228
223. Network Edge Prediction: Estimating the prior
Antoine Lizee, Daniel Himmelstein
ThinkLab (2016-04-14) https://doi.org/10.15363/thinklab.d201
224. Network Edge Prediction: how to deal with self-testing
Antoine Lizee, Daniel Himmelstein
ThinkLab (2016-04-05) https://doi.org/10.15363/thinklab.d194
225. Cataloging drug–disease therapies in the ClinicalTrials.gov database
ThinkLab (2016-05-08) https://doi.org/10.15363/thinklab.d212
226. A standard database for drug repositioning
Adam S. Brown, Chirag J. Patel
Scientific Data (2017-03-14) https://doi.org/10.1038/sdata.2017.29
227. Systematic analyses of drugs and disease indications in RepurposeDB reveal pharmacological, biological and epidemiological factors influencing drug repositioning
Khader Shameer, Benjamin S. Glicksberg, Rachel Hodos, Kipp W. Johnson, Marcus A. Badgeley, Ben Readhead, Max S. Tomlinson, Timothy O’Connor, Riccardo Miotto, Brian A. Kidd, … Joel T. Dudley
Briefings in Bioinformatics (2017-02-15) https://doi.org/10.1093/bib/bbw136
228. Toward a comprehensive drug ontology: extraction of drug-indication relations from diverse information sources
Mark E Sharp
Journal of Biomedical Semantics (2017-01-10) https://doi.org/10.1186/s13326-016-0110-0
229. Rephetio: Repurposing drugs on a hetnet [proposal]
Daniel Himmelstein, Antoine Lizee, Chrissy Hessler, Leo Brueggeman, Sabrina Chen, Dexter Hadley, Ari Green, Pouya Khankhanian, Sergio Baranzini
ThinkLab (2015-01-12) https://doi.org/10.15363/thinklab.a5
230. Measuring user contribution and content creation
Daniel Himmelstein, Antoine Lizee
ThinkLab (2016-04-11) https://doi.org/10.15363/thinklab.d200
231. This revolution will be digitized: online tools for radical collaboration
C. Patil, V. Siegel
Disease Models & Mechanisms (2009-04-30) https://doi.org/10.1242/dmm.003285
232. Publishing the research process
Daniel Mietchen, Ross Mounce, Lyubomir Penev
Research Ideas and Outcomes (2015-12-17) https://doi.org/10.3897/rio.1.e7547
233. Does it take too long to publish research?
Nature (2016-02-10) https://doi.org/10.1038/530148a
234. Accelerating scientific publication in biology
Ronald D. Vale
Proceedings of the National Academy of Sciences (2015-10-27) https://doi.org/10.1073/pnas.1511912112
235. Reproducibility: A tragedy of errors
David B. Allison, Andrew W. Brown, Brandon J. George, Kathryn A. Kaiser
Nature (2016-02-03) https://doi.org/10.1038/530027a
236. Workshop to analyze LINCS data for the Systems Pharmacology course at UCSF
Daniel Himmelstein, Kathleen Keough, Misha Vysotskiy, Jeffrey Kim, Beau Norgeot, Julia Cluceru, Marjorie Imperial, Emmalyn Chen, Jasleen Sodhi, Elizabeth Levy
ThinkLab (2016-03-08) https://doi.org/10.15363/thinklab.d181
237. Why we are teaching science wrong, and how to make it right
M. Mitchell Waldrop
Nature (2015-07-15) https://doi.org/10.1038/523272a
238. Going paperless: The digital lab
Nature (2012-01-25) https://doi.org/10.1038/481430a
239. Systematic integration of biomedical knowledge prioritizes drugs for repurposing
Daniel S. Himmelstein, Antoine Lizee, Christine Hessler, Leo Brueggeman, Sabrina L. Chen, Dexter Hadley, Ari Green, Pouya Khankhanian, Sergio E. Baranzini
Cold Spring Harbor Laboratory (2016-11-14) https://doi.org/10.1101/087619
240. Rephetio: Repurposing drugs on a hetnet [report]
Daniel Himmelstein, Antoine Lizee, Chrissy Hessler, Leo Brueggeman, Sabrina Chen, Dexter Hadley, Ari Green, Pouya Khankhanian, Sergio Baranzini
ThinkLab (2016-11-13) https://doi.org/10.15363/thinklab.a7
241. Figshare depositions from Project Rephetio
Daniel Himmelstein, Leo Brueggeman, Sergio Baranzini
Figshare (2017) https://doi.org/10.6084/m9.figshare.c.2861359.v1