How does flashpca handle missing data
51 views
Skip to first unread message
Nicole Torosin
Dec 9, 2018, 5:42:48 PM12/9/18
to flashpca-users
Hello,
I was just curious how flashpca handles missing genotypes. Does a sample with a missing genotype at a certain site contribute to the variance? Or is it ignored.
Thank you,
Nicole
Gad Abraham
Dec 10, 2018, 8:46:34 PM12/10/18
to Nicole Torosin, flashpca-users
Hi Nicole,
Missing genotypes are imputed to the average for that SNP, which after
standardisation becomes zero. The average and sd are based on the
non-missing genotypes for that SNP, so a sample with a missing
phenotype does not contribute.
If you have SNPs with high missingness, you should probably exclude
them beforehand using plink.
Cheers
Gad
> --
> You received this message because you are subscribed to the Google Groups "flashpca-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to flashpca-user...@googlegroups.com.
> To post to this group, send email to flashpc...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/flashpca-users/d684139e-e957-41ba-8673-3499b015d9c3%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
Missing genotypes are imputed to the average for that SNP, which after
standardisation becomes zero. The average and sd are based on the
non-missing genotypes for that SNP, so a sample with a missing
phenotype does not contribute.
If you have SNPs with high missingness, you should probably exclude
them beforehand using plink.
Cheers
Gad
> You received this message because you are subscribed to the Google Groups "flashpca-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to flashpca-user...@googlegroups.com.
> To post to this group, send email to flashpc...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/flashpca-users/d684139e-e957-41ba-8673-3499b015d9c3%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
Nicole Torosin
Dec 21, 2018, 1:26:49 PM12/21/18
to flashpca-users
Thank you so much.
On a related note, if I am evaluating X chromosome data using males and females, how does the PCA handle a combination of diploid and haploid genotypes?
Would you recommend using only females for this type of analysis?
Thank you again,
Nicole
Gad Abraham
Jan 2, 2019, 7:39:49 AM1/2/19
to Nicole Torosin, flashpca-users
Hi Nicole,
It really depends on what you're trying to do; if you want to detect
population structure or other large-scale variation like that, then
you don't need chr X, and the autosomal chromosomes will suffice
(assuming it's human). If you really want to do an analysis of chr X,
simply putting chr X genotypes in the PCA will probably "work" (not
crash) but may give garbage results, but I don't have an answer off
the top of my head how to do it in a sensible way that accounts for
sex.
Cheers
Gad
> To view this discussion on the web, visit https://groups.google.com/d/msgid/flashpca-users/a968749f-067c-4e47-934c-0b08f561b7a1%40googlegroups.com.
It really depends on what you're trying to do; if you want to detect
population structure or other large-scale variation like that, then
you don't need chr X, and the autosomal chromosomes will suffice
(assuming it's human). If you really want to do an analysis of chr X,
simply putting chr X genotypes in the PCA will probably "work" (not
crash) but may give garbage results, but I don't have an answer off
the top of my head how to do it in a sensible way that accounts for
sex.
Cheers
Gad
Reply all
Reply to author
Forward
0 new messages