abbvie-computationalgenomics/genetic-evidence-approval

genetic-evidence-approval's Issues

target_indication.tsv: First Added column documentation

target_indication.tsv includes a First.Added column. Here's the first rows for context:

MSH	ensembl_id	lApprovedUS.EU	Phase.Latest	First.Added	Inactive	symbol
Abortion, Spontaneous	ENSG00000119535	FALSE	unknown	NA	TRUE	CSF3R
Acidosis	ENSG00000151617	FALSE	Phase I Clinical Trial	2012-10-27	FALSE	EDNRA
Acne Vulgaris	ENSG00000012779	FALSE	Phase II Clinical Trial	NA	TRUE	ALOX5
Acne Vulgaris	ENSG00000026508	FALSE	Approved	NA	FALSE	CD44
Acne Vulgaris	ENSG00000057252	FALSE	unknown	2009-07-01	TRUE	SOAT1

The README documents this as:

First Added = earliest added date for any drug with this target and indication.

Is this the same metric that the manuscript refers to as "Earliest date in the Pharmaprojects event history" or "first recorded drug event"? This was my initial understanding, but the readme makes it sounds like First.Added might be when the Pharmaprojects drug record was created rather than when the first event for a drug is dated.

BTW thanks for the documentation and code & data sharing that accompanying this study!

Mapping Informa target families to GO terms

Hello, thank you for sharing this codebase!

In attempting to understand the derivation of certain variables used in the stan model, I came across this description of the process for generating target family indicator fields (in supplement 5):

GO Annotation Previous work has shown differences in success probability among target classes (e.g. kinases and GPCR) [6]. Pharmaprojects provides functional annotations which seemed to be predictive of success, but were not available for all targets. We translated these annotations to Gene Ontology (GO) [1][2] terms when possible, found all descendants of these terms using the goatools Python module [4], and then annotated each gene target with a term if it was annotated with that term or a descendant term. GO term annotations were obtained using biomaRt [7]. S21 Fig shows variation in approval and GWAS links by target class.

Is the mapping of PharmaProjects target annotations to GO terms used for this available anywhere?

Thanks.