This page provides answers to a wide array of general questions. For detailed tutorials on how to use DGIdb, see the tutorials under the "Getting started" section of the help menu.
What is the goal of DGIdb?
To help you annotate your genes of interest with respect to known drug-gene interactions and potential druggability.

How do I cite DGIdb?
Please cite: 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. DGIdb 2.0: mining clinically relevant drug-gene interactions. Nucleic Acids Research. 2016 Jan 4;44(D1):D1036-44. doi:10.1093/nar/gkv1165.

Is DGIdb open source and open access?
The source code for DGIdb is open-source and made available under the GNU General Public License version 3. The license is distributed with the source code (DBIdb license). The data used in DGIdb is all open access and where possible made available as raw data dumps in the downloads section.

How can I learn more about how to use DGIdb?
You can try reading: the 'Getting Started' section under the help menu, the manuscript, the tutorial, and this FAQ. Some sections also contain a 'Tour' that can be activated by pressing the 'Show Tour' button in the top left. If all else fails please email us with your questions. Refer to the 'Contact' section under the help menu for details on how to get in touch with us.

What types of gene names/symbols should I use as input?
The short answer is that you should ideally use official HUGO gene symbols as reported by Entrez Gene. We have developed a gene name grouping strategy that attempts to aggregate synonyms for each gene. We start with the official name and alternate names of each Entrez Gene and update with additional names from each data source. You should get reasonable results with known synonyms, Ensembl Ids, Uniprot Ids, etc. If you enter a gene name that is ambiguous or unmatched this will be noted in the results. You should attempt to find a more official name for such genes and update your gene list.

How are genes defined in DGIdb?
The gene summary page shows the primary name, alternate names and meta-data for each gene locus as provided by each source of data imported into DGIdb. Genes in DGIdb are first defined by Entrez but mapped together with gene records from other sources by Entrez ID, Ensembl ID, UniProt ID or alternate names/synonyms in that order of preference.

Where can I find the source code for DGIdb?
The code for DGIdb is available at github. Refer to the Downloads section for more details.

I have a source of druggable genes or drug-gene interactions that I would like to add to DGIdb. How do I do that?
This is an open source project and you could therefore create your own instance with your own data ... but if you want that source we probably do also. Contact us and we will be happy to consider adding your source for you.

What does 'Source Trust Level' mean?
We assign each source used by DGIdb to one of two trust levels: 'Expert curated' or 'Non-curated'. This status is internal to DGIdb. 'Expert curated' sources were examined manually by domain experts during the process of importing to DGIdb. 'Non-curated' sources were imported in an automatic fashion from the primary source using an importer module (i.e. a script). 'Non-curated' sources may still have involved some manual curation at the original source. Within each of the two trust levels we further rank each source according to our own assessment of quality. Results will be sorted to favor 'Expert curated' data and by quality rank within those sources.

What is the meaning of 'Gene Categories' in DGIdb?
Gene categories in DGIdb refers to a set of genes belonging to a group that is deemed to be potentially druggable. For example, kinases are generally deemed to have high potential value for development of targeted drugs. For more details on the sources of druggable gene category definitions, refer to the sources and background reading.

What is an 'Interaction Type'?
An interaction type describes the nature of the association between a particular gene and drug. For example, TTD reports the drug-gene interaction, SUNITINIB-FLT3. The interaction type is reported as, 'inhibitor'. Interaction type, as used in DGIdb is very broad. Dozens of interaction types are currently defined. Many interaction types describe the mechanism of action between a small molecule and a protein. Other broader types of 'interaction' might be used. e.g. Gene X mediates 'resistance' or 'sensitivity' to drug Y.

What is the difference between a drug-gene interaction and druggable gene category?
A drug-gene interaction is a known interaction (e.g. inhibition) between a known drug compound (e.g. lapatinib) and a target gene (e.g. EGFR). A druggable gene category is a grouping of genes that are thought to be potentially druggable by various methods of prediction. For example, the 'rule-of-five' analysis described by Hopkins and Groom in 2002 defined lists of genes that were most likely to be successfully targeted by small molecule inhibitors. Genes in these categories are *potentially druggable* and may or may not have existing drugs that target them. On the other hand, all of the genes from the drug-gene interaction sources are targeted by specific known compounds.

How is this application different from DrugBank or TTD?
There are many differences. DGIdb is limited to human genes only. DrugBank and TTD are databases that catalogue drugs and store detailed information about those drugs and the genes they target. DGIdb aggregates many such databases into a common framework. DGIdb adds considerable functionality for efficiently searching a list of input genes against these sources. DGIdb integrates both known drug-gene interactions and potentially druggable gene data. DGIdb allows the user to refine their query to certain gene families, types of interactions, classes of drugs, etc. DGIdb is open source and available in a format that would allow you to create your own instance. DGIdb incorporates sources of drug-gene interaction data that were previously only available in inaccessible formats (e.g. tables in a PDF document). DGIdb is meant to be used in combination with the original sources of raw data. Wherever possible we link out to those original sources.

I found a gene in a category that does not really belong there or a drug-gene interaction that appears false. Should I report it?
We are working on updates to the interface that would allows us to readily capture this kind of information directly from users. In the meantime please feel free to report it to us. The quality of our data is only as good as the sources of raw data. Throughout the results you will be able to see if multiple independant sources report the same findings. Using this as a filter may help you disqualify some spurious entries and have increased confidence in others.

Some druggable gene categories seem very inclusive ... for example are there really >800 human kinases?
This is by design. Druggable gene categories are aggregated from multiple sources, some of which are predictive in nature. Predictions based on sequence similarity etc. may result in some false positives. More stringent lists can be obtained by limiting to a particular source(s) (e.g. dGene only) or by requiring that multiple sources agree.

What does the 'interactions for genes related to $GENE' button on the genes pages do?
This button will submit a search for drug-gene interactions of genes that have direct gene-gene interactions with the current gene. For example, say you search for drugs that interact with the gene MPL. This search returns no drug interactions, but there are drugs that inhibit other genes that form complexes with or are in the same pathway as MPL. This button search for interactions with all of these genes. Gene-gene interaction information is obtained from Entrez Gene Interactions. Entrez obtains these interactions from BIND, BioGRID, HPRD, etc.

How is DGIdb implemented?
The web interface uses Ruby on Rails and Twitter bootstrap. The backend is a Postgres database. The importers that process data from each source and populate this database are written in object oriented Perl.

How is versioning handled in DGIdb?
This question has several answers. The footer of every page in the web interface contains a version stamp that describes versions for various components. The interface code itself is meticulously tracked in dgi-db at github. The source stamp contains a SHA-1 tag that corresponds to the version of code being run. Each input source of data we import may itself have a version. If so, we maintain this info and display it on the source summary view. If an input source does not have concept of versioning we note the date of import instead. Source version are included in TSV export results for each drug-gene interaction found.

What does the version stamp at the foot of each view mean?
Its currently generated whenever we perform a rake task to create a new data snapshot for DGIdb. The version number corresponds to a tag in git and eventually the debian package we use to push a development version of DGIdb to the publicly facing server.. The SHA-1 is the commit id from when the snapshot was generated, as is the datestamp.

I think I have identified a bug in the DGIdb code. What should I do?
Please let us know. For minor issues, use the feedback form in the bottom right of views at dgidb.org. If you question is more complicated, please ask it publicly on BioStars. Finally, the code for DGIdb is open-source, if you are the curious/ambitious type, feel free to investigate a solution and let us know what you find.