Closed jainy closed 2 years ago
Hi Jainy, since it's folly for me to try to keep up with all possible builds and annotations, slivar make-gnotate
is available so you can make your own gnotate/zip file for any VCF. I think the spliceAI would require some preprocessing (I made a new field that was the max of the 4 scores, I think). I'm happy to help with any difficulties you encounter using make-gnotate
.
Thank you for the quick reply and pointers, Brent! I added the max of the 4 scores to the vcf and ran make-gnotate
. It worked fine.
Excellent! Glad to hear it.
In SpliceAI v1.3 spliceai_scores.masked.snv.hg38.vcf.gz
, some variants were duplicated with different max scores. Would you mind sharing how these should be dealt with in preprocessing after taking max of the 4 values in a row?
For the example below, would slivar give 0.02 or 0.09 or both for 12-121527653-C-G?
12 121527653 . C A . . SpliceAI=A|KDM2B|0.00|0.00|0.00|0.00|-3|-27|-3|39
12 121527653 . C A . . SpliceAI=A|KDM2B|0.03|0.00|0.00|0.00|-3|-27|39|7
12 121527653 . C G . . SpliceAI=G|KDM2B|0.02|0.00|0.00|0.00|-3|-9|-3|7
12 121527653 . C G . . SpliceAI=G|KDM2B|0.09|0.00|0.00|0.00|-3|-1|39|7
12 121527653 . C T . . SpliceAI=T|KDM2B|0.00|0.00|0.00|0.00|-3|-1|-3|0
12 121527653 . C T . . SpliceAI=T|KDM2B|0.00|0.00|0.00|0.00|-3|-9|-3|39
I looked over the code and it looks like both values will be stored and likely only the first (after unstable sort, so either could be first) will be reported on annotation, so if you want to control this, you'd have to collapse records like these yourself before sending to make-gnotate.
Hi Brent,
I see that spliceai scores available are for hg37. It will be great if you provide for the hg38 version as well. Thank you!
Best, Jainy