Bug in top gene selection?

phenotips / remote-matching

Patient matching across applications

2 stars 2 forks source link

Bug in top gene selection? #16

Closed buske closed 9 years ago

buske commented 9 years ago

For some reason, the top gene included for matchmaking from F0000010 is ATP6V0A4, rather than SRCAP, which actually has a higher score in the exomiser-annotated VCF file. ATP6V0A4 is the second-highest.

allasm commented 9 years ago

I will also double-check how top genes are selected (current code gets the list of all genes and computes all scores and picks top N outside of exomiser)

buske commented 9 years ago

Yeah, I took a look at that code and couldn't find any obvious issue. Could be a problem with patient-network, but the gene does show up correctly in internal patient matching.

buske commented 9 years ago

Found the bug. It's actually returning the _last_ (lowest-ranked) gene in the file, due to ascending rather than descending sort order. It's just the second and last gene names were similar.