High overdispersion with autopolyploid potato data

[heatherturtle]

I have over 450 individuals that are tetraploid and I used the GATK pipeline. I have the vcf containing diploid calls that was filtered on missingness per individual, missingness per site, depth, FS, GQ, ReadPOsRankSum, and MQ. When I follow the polyRad workflow, it says that overdispersion is optimal at 30. However, after running ExpectedHindHe it leads me to results that remove over 95% of my markers (out of ~16K). My thoughts are that I am filtering to stringently, but I would like your input on the results that I am getting.

Thanks!

[lvclark]

Could you share your histogram of Hind/He values for markers with MAF > 0.05, prior to any filtering by Hind/He?

[heatherturtle]

Hi Lindsay, Thanks for the quick response! Histogram is attached. Best, Heather

…

On Thu, Feb 17, 2022 at 12:33 PM Lindsay Clark ***@***.***> wrote: Could you share your histogram of Hind/He values for markers with MAF > 0.05, prior to any filtering by Hind/He? — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMURL2MYJJURH7OD357YPRDU3U5O5ANCNFSM5OS44Q5Q> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[lvclark]

Hi Heather,

I don't see the histogram. Could you upload it directly on GitHub rather than attaching it to email?

Thanks,

Lindsay

[heatherturtle]

[lvclark]

Ok, from this it looks like you should set inbreeding to about 0.3 when using ExpectedHindHe. Is that the value that you used?

[heatherturtle]

That is not the value I used! It was set to zero, so it was a typo! I am only removing 30% of markers now. In an autopolyploid, I would only want to remove markers where Hind/He is too low, correct?

[lvclark]

I would recommend also removing markers where Hind/He is too high, even though you aren't working with an allopolyploid. If Hind/He is too high, it could indicate misalignment or sequencing error.

Glad this helped!

[heatherturtle]

One more thing I want to be clear on, is that polyrad is intended to impute tetraploid genotype calls from previously diploidized genotype calls, is that correct? Is it beneficial to use tetraploid calls (GATK is able to produce tetraploid calls) before using polyRAD, or is it fine to keep them as diploid calls? Thank you for taking time out of your day to answer these questions, it is greatly appreciated!

…

On Fri, Feb 18, 2022 at 3:13 PM Lindsay Clark ***@***.***> wrote: I would recommend also removing markers where Hind/He is too high, even though you aren't working with an allopolyploid. If Hind/He is too high, it could indicate misalignment or sequencing error. Glad this helped! — Reply to this email directly, view it on GitHub <#26 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMURL2O7AORHV4RU5FP4KXLU32ZBBANCNFSM5OS44Q5Q> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[lvclark]

That's actually not correct. When polyRAD imports from VCF, it ignores the genotype calls entirely. The allelic read depths from the VCF are used by polyRAD to make new genotype calls.

Still, I can see it being advantageous to do tetraploid genotype calling in GATK if you know your organism is tetraploid, because that may impact some of the filtering parameters.