QC UKB

[Maria Di Biase]

Dear Christopher,

I would like to filter out SNPs as follows:

1.     with a minor allele frequency (MAF) < 0.01, 

2.     with a Hardy-Weinberg equilibrium (HWE) test p value < 10−7 

3.     with a proportion of missingness (Pm) > 0.05

4.     with an imputation information score < 0.8, 

5.     that are a duplicated SNP

I am using ukbiobank data and am new to PLINK. Can I achieve steps 1-3 as follows?

plink2 --bgen ukb_imp_chr1_v3.bgen ref-first --sample ukb22828_c1_b0_v3_s487253.sample --maf .01 --hwe 1e-7 --mind .05 --make-bpgen --out plink_outputs/chr1 

How can I achieve steps 4 and 5? Can this be done in the same command or is a separate command needed?

Thank you kindly in advance,

Maria

[Christopher Chang]

Yes, that command should handle steps 1-3.

--mach-r2-filter can be used to filter on the MaCH-r2 imputation statistic in the same command; alternatively, you can use --extract-col-cond to filter on another imputation statistic in a text file.

--rm-dup can be used to filter duplicate SNPs.  I recommend performing this step separately, so you can first look at what sort of duplicate SNPs you have and then decide how you want to handle them.

[Maria Di Biase]

Thank you very much for the quick turn around. 

Re the imputation score, I would like to use column 7 in a file called "ukb_mfi_chr1_v3.txt" to remove SNPs with an info score < 0.8. Column 1 of this file specifies the marker ID (the marker order is not necessarily  the same as the BGEN files). Is the following use correct?

--extract-col-cond-min .8

 ukb_mfi_chr1_v3.txt 7 1 

Also, is there a command to characterize the sort of duplicate SNPs that are present in the data? 

Thanks again,

Maria

[Christopher Chang]

No, "--extract-col-cond ukb_mfi_chr1_v3.txt 7 1 --extract-col-cond-min 0.8" is probably what you want.

--rm-dup's default mode ("error") checks groups of duplicate-ID SNPs for equality; if any mismatches are found, it errors out and reports SNPs with mismatches.  You can then decide what to do about the mismatching same-ID SNPs.