SuLab / GeneWikiCentral

GeneWiki Organization
MIT License
5 stars 2 forks source link

Drug Indications #87

Open stuppie opened 6 years ago

stuppie commented 6 years ago

Notes regarding Drug Indications:

Currently in Wikidata, we have the property "drug used for treatment", which was mostly populated with the "may_treat" statements from NDFRT (Example), but also data based on in vitro experiments and clinical trials from CHEMBL (example: https://www.ebi.ac.uk/chembl/compound/inspect/CHEMBL1064). The later caused problems and (I think) were removed. The data from NDFRT is not very accurate or up to date either. (Example: ALATROFLOXACIN says it may treat infections in NDFRT, but it was pulled from the market in 2006)

We also have therapeutic area which is a less specific "disease area in which a medical intervention is applied". These are populated with disease areas from EMA (Example), and have dates associated, so drug removed from market are not represented. The therapeutic areas are mostly diseases, but not always (i.e. smoking cessation)

The property "may prevent" was recently added. Currently unused.

Potential Data Sources:

Resource name Example record Structured? Open license? Continuously updated? Last updated Provenance? Notes?
Drugbank  Fomepizole Diseases structured using drugbank IDs, not linked to anything else. commercial version includes a text description, severity, type of indication, and associated ICD10 or MedDRA identifiers  no FDA drug labels and scientific publications    
Drugcentral  Lipitor   CC-BY-SA they claim so, but no proof yet     For drugs approved after 2012 disease annotations are extracted manually from approved drug labels and mapped to SNOMED-CT concepts. Targets are from chembl and WOMBAT-PK
NDF-RT     yes     no  
Clue.io All records  Not really   ?      "Surprisingly, we found that it was difficult to determine even the approved clinical indications for existing drugs. The availability of such information in public databases is highly variable, and in many cases, terminology is inconsistent. Using the NIH's DailyMed repository of FDA-approved drug-prescribing labels and other resources, we manually curated a dictionary of 644 unique drug-indication terms and assigned them to 1,918 launched drugs" paper
FAERS/AEOLUS data yes probably yes     This is a database of drugs and what they were used for. Does not distinguish between types of therapies, does not distinguish between approved and therapeutic usages. For example, aspirin -> Myocardial infarction
LabeledIn       no      himmelstein thinklab code
MEDI--an Ensemble MEDication Indication Resource   yes yes no      
ChEMBL CHEMBL941 yes CC-BY-SA yes   not enough doesn't adequately distinguish investigational uses from approved uses. Could be inferred by presence of a dailymed reference though...
pharmacotherapydb   yes CC-0 no     dhimmel's resource, incomplete
orphanet Miltefosine sort of Data Transfer Agreement for Academia yes ?   Looks like data comes from here: http://www.orpha.net/orphacom/cahiers/docs/GB/list_of_orphan_drugs_in_europe.pdf
EMA palbociclib semi-structured CC-0 yes   yes, but not open, drug study documents from sponsor, containing academic references, accessible upon registration  
dailymed Ziagen semi-structured   yes      
clinicaltrials.gov   semi-structured CC-0 yes   if results are being reported and published, reference is provided Many studies just completely lack any data except whether a study is active and recurting patients, because that's what study sponsors need to provide by law.
SPL-X (structured product label extractor) yes   no      
Wikipedia/infoboxes Gout sometimes yes yes sometimes