Higher heterozygosity than expected

[zlye]

My results are giving higher heterozygosity that I expected. I am working with a selfing plant species (rice). I am genotyping with the population genotyper method in the lumpy pipeline (svtyper). All my samples have >10x sequencing coverage. As far as I can tell svtyper was developed for use in humans, and therefore has a prior set for human rates of heterozygosity. Is there any recommendation for how to use svtyper with inbred individuals? Is it it possible to change the prior? I am planning to assume all heterozygous calls to be alternates but was curious if you'd have any suggestions for this problem.

[brentp]

hi, I would first try to filter the (DUP and DEL) results using the duphold annotations, requring that any sample that is het for a DEL also has DHFFC < 0.7 and for DUP, DHFFC > 1.2. Then see how it looks.

[zlye]

Hi Brent! thanks for your speedy response - I've already applied those filters as recommended. Do you have any other tips? Cheers

On Mon, Nov 18, 2019 at 12:31 PM Brent Pedersen notifications@github.com wrote:

hi, I would first try to filter the (DUP and DEL) results using the duphold annotations, requring that any sample that is het for a DEL also has DHFFC < 0.7 and for DUP, DHFFC > 1.2. Then see how it looks.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_brentp_smoove_issues_95-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DAKFUYUYXD36DXCOXYXMOSDTQULGOVA5CNFSM4JOWVF4KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEELICZY-23issuecomment-2D555123047&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=163oVAfqYW8rLwmeRfPScQ&m=HZqV7sS5UnIRk-1AOr7VMV_T91Cdi_ibq_PYNcdVakk&s=QcVZ9MKaD54DxNNgTPl0iOdixyiIM1gSGyXOjlTkWLw&e=, or unsubscribe https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AKFUYU6ENNTDASAZYZ6CPZDQULGOVANCNFSM4JOWVF4A&d=DwMCaQ&c=slrrB7dE8n7gBJbeO0g-IQ&r=163oVAfqYW8rLwmeRfPScQ&m=HZqV7sS5UnIRk-1AOr7VMV_T91Cdi_ibq_PYNcdVakk&s=Gb4am9_H5cPa4b-llVFaqfbAaEqpZfgl0xmIoO-gWio&e= .

-- Zoe Lye P.h.D. candidate 12 Waverly Place New York University New York, NY 10003

[brentp]

the duphold filters are the best available. i guess you could also limit to stuff larger than 300 bases and be a bit more stringent for duplications.

for deletions > 300 bases, with duphold support, the calls should be quite good.