search

griffithlab / civic-server

Backend Server for CIViC Project
MIT License
39 stars 32 forks source link

Possible mismatch between Illumina data and our variants #307

Open kkrysiak opened 7 years ago

kkrysiak commented 7 years ago

We should dig into this a little more and how this matching occurred. I feel like something went wrong here but I need more time to determine whether this was due to our coordinates or due to something in the Illumina data.

https://civic.genome.wustl.edu/#/events/genes/5/summary/variants/563/talk/revisions/list/13134/summary#variant

acoffman commented 7 years ago

So looking at the illumina data, this matched on both beMarkerId (V600K) and coordinates 7:140453136-140453137 AC -> TT. For this variant they have an rsid of rs113488022 and aminoAcidChange of V600fs which is where these aliases are coming from.

Their dataset has an additional variant (same coordinates except the variant bases are AC -> CT) which has the same rsid and aminoAcidChange but a different beMarkerId of V600R.

kkrysiak commented 7 years ago

Several of their rsIDs for indels are just wrong. Particularly for KIT. We may want to adjust our approach moving forward for the rsID matching but I'm not sure...

kkrysiak commented 7 years ago

We are having a different issue with some of the EGFR variants. Particularly L747_A750>P has two different sets of aliases that were added.

acoffman commented 7 years ago

For https://civic.genome.wustl.edu/events/genes/19/summary/variants/1006/talk/revisions/list/15041/summary#variant

The coordinates don't appear to line up, nor does the aminoAcidChange, but these matched up on beMarkerID (L747_A750>P).

kkrysiak commented 7 years ago

I've also noticed that we've been adding a "-" for inframe deletions but in other places these are blank in the interface. Is this a problem or something we should change next time around?