Remove Chr:position format to just Chr

Question

[kstafford32]

Hello all,

I attempted to run a PRS analysis and received the error:

84489606 variant(s) not found in previous data
0 variant(s) included

Error: No vairant remained!

I think it is because of the differing format between my base and target data:

##BIM TARGET
22      chr22:10510227:A:G      0       10510227        G       A

##BASE MDD.QC.gz
rs      ref     alt     pval    effref  info    chr     pos     reffrq  N
rs10    A       C       0.9576  -5e-04  1       7       92383888        0.0596  500199

How can I remove the chr:position:allele format in the target bim file? I believe I need awk print but not entirely sure how to produce this code. Google was no help.

I wonder if the rs snp format in the base data is also causing this error. If so, how can I add the rs format to my target data or remove the rs format from my base data.

Thank you immensely!

[williamsbrian5064]

Do you want to remove the chr:position:allele format completely?

Answer 1

[williamsbrian5064]

You could use the following on the target bim file and see if that helps. It should remove the column completely from the target bim file

awk '{$2=""; print $0}' target.bim > new.target.bim

[kstafford32]

Thank you, William. Would this move it completely? Can I keep the chr22 and just remove the following position/allele info?

Also, any idea of how to update my snpid with rs in my target file?

[williamsbrian5064]

Do your target file and base data have the exact same positions?

[rpolicastro]

That might still leave the column there, just with empty contents. If you want to remove the second column you can use cut cut -f-2 target.bim > new.target.bim. You can adjust the delimiter with -d.

[williamsbrian5064]

You could also do this if you wanted to keep chr22

awk '{sub(/:.*/,"",$2)}1' OFS='\t' target.bim > new.target.bim

[cpad0112]

Please add tab as OFS. Is it 6 or 1?

[williamsbrian5064]

It should be 1. Sorry, that was my mistake

Answer 2

[cpad0112]

with sed:

$ echo -e '22\tchr22:10510227:A:G\t0\t10510227\tG\tA'
22      chr22:10510227:A:G      0       10510227        G       A

$ echo -e '22\tchr22:10510227:A:G\t0\t10510227\tG\tA' | sed -r 's/:[0-9]+:[A-Z]+:[A-Z]+//'
22      chr22   0       10510227        G       A

Answer 3

[Sam]

Based on your log, it seems like you are using PRSice. You can use --chr-id C:L with the latest version (2.3.3), which should automatically do the chr ID generation for you.

Edit: just realize it is you. Use --chr-id should solve the problem.

[kstafford32]

Thank you, Sam. So, for my target file, it looks like my snps are listed in this format: chr1:17847:T:C, with chr#:position:a1:a2. I have no idea if it's supposed to be formatted this way. My base file snp's are formatted as rs, and I've specified this in the script by using --snp rs, as you've previously suggested.

Would --chr-id C:L work if my bim file does not have headers? Know of a quick way to add headers to my bim if needed? Lastly, does PRSice recgonize this --chr-id C:L:a1:a2 ?... is it necessary to format it this way considering my format is chr1:17847:T:C in the target file?

Thank you!

[Sam]

Because bim file is well defined and we know that it always contain the required information in specific columns. And definitely don't manually modify the bim file, as that breaks the format guideline.

--chr-id C:L works because setting that ask PRSice to ignore the SNP column in the base file, and try to construct a new ID using the chromosome (C) and base pair (L). You can get the a1 a2 using a and b respectively. So something like --chr-id C:L:a:b will do. However, if you can run your script without a and b, I will definitely do that, because by adding the A1 and A2 in your SNP ID, you effectively forbid PRSice from performing strand flipping, which can lead to information lost.