This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Exclude non shared SNPs from two datasets with PLINK

Hello,

I have two SNPs datasets that I need to merge. I have created a file called "shared.bim" which contains all the shared sites and a reference_allele.list to reorder the sites in both files before merging.

Since trying to merge without first removing the non shared sites would give an error ("Warning - impossible allele assignment") I need to remove the non shared sites from both datasets.

I know I need to create a command that contains --recode and --make bed and have a SNP-LIST-FILE to indicate what to remove; I just don't understand how to create the list of the SNP that I need to remove.

Is there a way to simply tell with a script: "remove those not contained in "shared.bim"?

Thank you

human snps genome shared plink sites
• 1,849 views • link
written by GIACOMO FRANCESCO • GIACOMO FRANCESCO

1 answer

plink is very well documented : https://www.cog-genomics.org/plink/1.9/filter#snp

use --extract command to extract SNP ID of your choice in both datasets prior to merge.

Also check the merge section : https://www.cog-genomics.org/plink/1.9/data#merge

• 0 views • link
written by Nicolas Rosewick Nicolas Rosewick

Hello,

I don't have problems with the merging process, just on how to flag the non shared ones

I did create a text file called "shared_sites.txt", which contains the shared sites taken from shared.bim.

I wrote this script to prepare for merging (I will merge in VCFtools):

plink --bfile dataset1 --recode vcf --a2-allele reference_allele.list --keep-clusters shared_sites.txt --out dataset1.filtered

I have put --keep-clusters shared_sites.txt to try flagging the shared sites to keep, while --a2-allele reference_allele.list is meant to reorder the snps according to the list.

Is the keep cluster command going to work for this purpose?

• 0 views • link
written by GIACOMO FRANCESCO • GIACOMO FRANCESCO

Log in to answer this question.

More posts like this