DGIdb 2.0: mining clinically relevant drug-gene interactions.
Wagner AH, Coffman AC, Ainscough BJ, Spies NC, Skidmore ZL, Campbell KM, Krysiak K, Pan D, McMichael JF, Eldred JM, Walker JR, Wilson RK, Mardis ER, Griffith M, Griffith OL. · 2016
Read DGIdb 2.0: mining clinically relevant drug-gene interactions. by Wagner AH, Coffman AC, Ainscough BJ, Spies NC, Skidmore ZL, Campbell KM, Krysiak K, Pan D, McMichael JF, Eldred JM, Walker JR, Wilson RK, Mardis ER, Griffith M, Griffith OL. (2016) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4702839.txt.
Introduction
Alex H. Wagner, Adam C. Coffman, Benjamin J. Ainscough, and nine colleagues released DGIdb 2.0 in Nucleic Acids Research in 2016. The Drug-Gene Interaction Database gathers claims about which drugs act on which genes, and which genes look druggable, from many scattered sources. Version two drew on twenty-seven sources after heavy hand curation. Nine new sources were added, seven of them tied to clinical trials, more than doubling the interaction count.
Claims about druggable genes rose by thirty percent. Most public sources behind the database now refresh automatically each week, keeping entries current. A new search by drug name joins the older gene-first search, and the programming interface was documented for other software. The result is a practical mining tool for precision medicine: clinicians and researchers can ask which known drugs touch a patient's gene of interest, and how strong the evidence is.
Catalog record
| CREATOR | Wagner AH, Coffman AC, Ainscough BJ, Spies NC, Skidmore ZL, Campbell KM, Krysiak K, Pan D, McMichael JF, Eldred JM, Walker JR, Wilson RK, Mardis ER, Griffith M, Griffith OL. |
|---|---|
| YEAR | 2016 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
DGIdb 2.0: mining clinically relevant drug–gene interactions By Alex H Wagner, Adam C Coffman, Benjamin J Ainscough, Nicholas C Spies, Zachary L Skidmore, Katie M Campbell, Kilannin Krysiak, Deng Pan, Joshua F McMichael, James M Eldred, Jason R Walker, Richard K Wilson Published in: Nucleic Acids Research DOI: 10.1093/nar/gkv1165 ABSTRACT Abstract The Drug–Gene Interaction Database (DGIdb, www.dgidb.org) is a web resource that consolidates disparate data sources describing drug–gene interactions and gene druggability. It provides an intuitive graphical user interface and a documented application programming interface (API) for querying these data. DGIdb was assembled through an extensive manual curation effort, reflecting the combined information of twenty-seven sources. For DGIdb 2.0, substantial updates have been made to increase content and improve its usefulness as a resource for mining clinically actionable drug targets. Specifically, nine new sources of drug–gene interactions have been added, including seven resources specifically focused on interactions linked to clinical trials. These additions have more than doubled the overall count of drug–gene interactions. The total number of druggable gene claims has also increased by 30%. Importantly, a majority of the unrestricted, publicly-accessible sources used in DGIdb are now automatically updated on a weekly basis, providing the most current information for these sources. Finally, a new web view and API have been developed to allow searching for interactions by drug identifiers to complement existing gene-based search functionality. With these updates, DGIdb represents a comprehensive and user friendly tool for mining the druggable genome for precision medicine hypothesis generation. INTRODUCTION With the increasing availability and decreasing cost of molecular profiling methods, growing attention has been paid to the use of these technologies for characterizing the mechanisms of human disease at the cohort—and more recently, the individual—level (1–3). Improved understanding of such molecular characteristics has been demonstrated to impact the diagnostic, prognostic, and treatment decisions made by clinicians, yielding positive results, less toxicity and improved quality of life (4,5). However, finding well-annotated drug–gene interactions relevant to medical decision-making remains an ongoing challenge. Several resources exist that characterize both established and experimental drug–gene interactions but these resources vary widely in data structure and complexity. Many of these databases are missing useful features provided by DGIdb, such as tools for querying lists of terms (6,7), filtering based on interaction type (6–9) or gene class (7–9), and a RESTful API (7–10). Others lack a search interface entirely, and instead have only a browsing interface (8) or are available only as static supplementary documents published online (11–20). DGIdb 1.0 (21) was first introduced as a novel resource to enable mining of multiple existing sources of drug–gene interactions and druggable gene categories. drug–gene interactions are the observed or inferred interaction between gene products (targets) and drug compounds (ligands) obtained by literature mining or by parsing publicly available databases. Many of these interactions include information on interaction types that specify how the drug interacts with the gene. The curation of DGIdb sources for interaction type data resulted in 33 distinct interaction type terms. In simple cases, these terms were harmonized where the interaction types were unquestionably equivalent (e.g. ‘inhibitor’ = ‘inhibitors’). However, in most cases we felt it best to preserve any subtleties of the interaction type according to the source. Druggable gene categories include genes that currently may not be targeted therapeutically but are potentially druggable according to their membership in gene categories associated with druggability (e.g. kinases) (21). These categories for the most part were defined by the original seminal paper by Hopkins and Groom (12) who first defined the ‘druggable genome’ as genes belonging to over 130 specific biological types (kinases, nuclear hormone receptors, etc.) that were thought to be druggable, and our curation of these types reduced that definition to 21 distinct types that were incorporated into DGIdb. Russ and Lampel (16) published an update of the ‘Druggable Genome’ concept but without breaking down genes into specific categories. The dGene resource (13) was the most recent update to the concept of a ‘Druggable Genome’ list. These lists, while widely cited and used, were not available in any conveniently usable form. Therefore, we created a generic ‘Druggable Genome’ category, comprised of the union of these three sources to act as a convenient label for these definitive lists. Other categories from other sources are more specific. The ‘Clinically Actionable’ category is a new concept with the DGIdb 2.0 release describing lists of genes that are actively being used in targeted clinical sequencing panels for precision medicine in cancer. The druggable gene categories of DGIdb are by no means mutually exclusive, and in fact often overlap extensively if not entirely (serine threonine kinases, for instance, are a subset of all kinases). Populating DGIdb 1.0 was accomplished through an extensive curation effort. This process typically included manually extracting and reviewing records, converting them to represent the concepts stored in the DGIdb, and writing custom importers to load the data provided by each source. Thirteen distinct sources were curated into a broad resource that is simple to query and explore. Druggable gene categories could be browsed without any structured query, and the interface accepted most common gene nomenclatures as identifiers for searching drug–gene interactions. Data describing interactions or potentially druggable categories (referred to as interaction or category claims) from each of these sources were linked to corresponding drug and gene concepts. Each such source typically included important metadata characterizing aspects of the interaction, such as if a particular gene confers resistance or sensitivity to the drug. The breadth, quality, and ease-of-access enabled by this resource is reflected in the substantial web traffic DGIdb receives, with ∼1500 unique users and ∼2300 sessions a month, with an average duration of ∼4.4 min/session. In addition, a number of other bioinformatics tools have used DGIdb in developing their own platforms including PANDA (22), iCAGES (icages.usc.edu), BioGPS (23), OmicsPipe (24), GEMINI (25), StationX (www.stationxinc.com) and IHLDB.rf (www.lungcancerdatabase.com), highlighting the broad usefulness of the DGIdb API. The first version of DGIdb was released in October 2013, and has since been cited in over 50 publications. Despite its popularity, DGIdb 1.0 was limited in several ways. The gene-centric search interface allowed the user to find known drug interactions for a set of gene targets, but did not provide a straightforward way of querying known gene targets for a set of drugs. In addition, several popular resources describing the druggable genome and drug–gene interactions were missing from DGIdb. Finally, some of the resources parsed by DGIdb are continually updated, creating a need for repeated manual updates to keep the associated data current. With this update, we address these shortcomings. DGIdb 2.0 includes 14 new sources added since version 1.0 for a cumulative total of 27 sources describing drug–gene interactions and druggable gene categories (Tables 1 and 2). These include nine new sources describing 10,102 drug–gene interactions potentially relevant to clinical actionability (e.g. relevant clinical trials, etc.). With this focus, DGIdb is better positioned as a resource for drug–gene interaction hypothesis generation in precision medicine efforts. By reviewing drug–gene interactions that are clinically relevant, basic researchers and physician-scientists using DGIdb have a more useful reference for hypothesizing which treatments might provide benefits for individual patients. Of special note are the addition of drug–gene interactions from the Clinical Interpretations of