Covariance Matrix Filled with NA

[emorywerner]

The uncentered.cov.mat generated by the SampleCovSNPs function is a matrix full of NA:

scaled.snp.by.pop looks like:

Is there a way to overcome this issue?

[jjberg2]

hmm, are any of the entries in scaled.snp.by.pop NA?

[emorywerner]

Yes there was. It is working once I removed them. Thank you.

[emorywerner]

Removing the NA's from scaled.snp.by.pop worked for some but not all of the files. Some of the files now are generating the matrix filled with NaN's. I tried removing NaN's from scaled.snp.by.pop with no luck.

[jjberg2]

Hmm, are any of the SNPs fixed across all populations? This would result in dividing by zero when rescaling frequencies, and thus, NaN. That’s my best guess. FYI I’m about to be without email contact for 4-5 days, but will answer your question about weighting the means when I get back.

On Wed, Aug 28, 2019 at 8:38 AM emorywerner ***@***.***> wrote: Removing the NA's from scaled.snp.by.pop worked for some but not all of the files. Some of the files now are generating the matrix filled with NaN's. I tried removing NaN's from scaled.snp.by.pop with no luck. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#1?email_source=notifications&email_token=AA67P6W5WKQDPV2SICIV7BDQG2EWXA5CNFSM4IPDGYB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5LLGJA#issuecomment-525775652>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA67P6V5BY5ZWZ3T6TH76BDQG2EWXANCNFSM4IPDGYBQ> .

-- -Jeremy

[emorywerner]

Hi Jeremy,

Thanks so much.

There aren't SNPs where the AF is constant for all the populations but there are SNPs where the AF is 0 for all populations except for one. However I am not sure how to deal with this. Should I exclude these SNPs? Not sure if these SNPs just do not exist in these populations or they have not been recorded in the populations.

Thanks!