svaba
svaba copied to clipboard
walaj
calculate Variant Allele Frequency (VAF)
Hello,
I was wondering what the correct way to calculate the VAF is? I tried AD/DP but a lot of the regions have a value greater than 1 and they also don't correspond with the VAF that some other callers are giving.
Thanks!
https://github.com/walaj/svaba/issues/29
I see you found issue #29 which describes why this is more difficult / less-well-defined for SVs than for SNVs. It's a good question. I would suggest that the best way to do this would be to use only split-reads (SR in the genotype field), since these reads fall directly on the breakpoint.
I see you found issue #29 which describes why this is more difficult / less-well-defined for SVs than for SNVs. It's a good question. I would suggest that the best way to do this would be to use only split-reads (SR in the genotype field), since these reads fall directly on the breakpoint.
hello, what if the SR==0 and AD/DP > 1, IS it indicated that the variatn is aritifical? if no, how could we calculate the correct VAF?
It seems that would mean that this was supported by only discordant reads. Depends on your library — if it is low-coverage large-insert size (like a mate pair library) then it’s probably good support. If it’s a more standard WGS library, it is suspicious.
Take a look at the mapping quality of the discordant reads, that could be a clue to the quality of the variant.
Without split reads, it’s hard to get a reliable VAF.
On Tue, Dec 13, 2022 at 9:41 PM dfc-1931475 @.***> wrote:
I see you found issue #29 https://github.com/walaj/svaba/issues/29 which describes why this is more difficult / less-well-defined for SVs than for SNVs. It's a good question. I would suggest that the best way to do this would be to use only split-reads (SR in the genotype field), since these reads fall directly on the breakpoint.
hello, what if the SR==0 and AD/DP > 1, IS it indicated that the variatn is aritifical? if no, how could we calculate the correct VAF?
— Reply to this email directly, view it on GitHub https://github.com/walaj/svaba/issues/50#issuecomment-1350289527, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABUZ7CCCNFROWMIOPSGMM7TWNEXV7ANCNFSM4GYI4XNQ . You are receiving this because you commented.Message ID: @.***>
Thank you for timing and kind reply.There is another question i wanna ask. Since my data is from targeted panel sequecing, An information occured , when a variants SR != 0, but its DP < SR, so SR/DP > 1, Is this a bug? It`s just something like below
.
BY THE WAY, IS the LO parameter in fusion-SV can be used to fileter the variant? Can i simply regard the LO IS BIGGER, THE variatnt is more real ? Thanks again for your great kindness!
This is more odd, but I suppose could be explained by the 2nd issue with SVs in that the read depth could be defined at either breakpoint A or breakpoint B. It also depends on whether the clipped portion of the read is included in the depth at that site.
It looks like I set the depth as the average of the two https://github.com/walaj/svaba/blob/e384f5c53913744d2baf822a199dc20ba0d0a4c6/src/svaba/BreakPoint.cpp#L1459
I haven't heard of this being a bug otherwise, and it looks like this is being called with high-confidence. May be some quirk with this particular case, but hard to say without knowing more.
Yes, the bigger the LO the more likely the data fit the variant model than the artifact model.
On Fri, Dec 16, 2022 at 3:52 AM dfc-1931475 @.***> wrote:
Thank you for timing and kind reply.There is another question i wanna ask. Since my data is from targeted panel sequecing, An information occured , when a variants SR != 0, but its DP < SR, so SR/DP > 1, Is this a bug? It`s just something like below [image: image] https://user-images.githubusercontent.com/55040570/208060002-abe4eed9-f5f7-46fd-b5db-43dcc7ca5710.png . BY THE WAY, IS the LO parameter in fusion-SV can be used to fileter the variant? Can i simply regard the LO IS BIGGER, THE variatnt is more real ? Thanks again for your great kindness!
— Reply to this email directly, view it on GitHub https://github.com/walaj/svaba/issues/50#issuecomment-1354405657, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABUZ7CCCGWL2GF5WK34VDK3WNQUUDANCNFSM4GYI4XNQ . You are receiving this because you commented.Message ID: @.***>