Hi All
I've tried to merge two data files (binary) with genome wide SNP data but I've got a few hundred warnings:
plink --bfile file1 --bmerge file2.bed file2.bim file2.fam --make-bed --out join2
Warning: Multiple positions seen for variant 'rsxxxxxx'.
Warning: Multiple chromosomes seen for variant 'rsxxxxxx'.
I've checked the chromosome numbers and the files do have different chromosome numbering. The SNP positions are also off by 1bp between SNPs with the same name. Both datasets should be h37 build so I'm not sure why the there are these differences and I don't know what I should do to reposition the SNPs nor onto which chromosome they should be repositioned.
What was the source of the data? Did you generate them or download them? I would be worried to adjust chromosome and position values because of the risk or erroneously merging SNPs together which actually reference different loci- unless you are 100% sure of the shift between the two files (e.g. in file2, chr1 is always labelled chr2), in which case you could do this for a large number of entries easily if you are familiar with any scripting language.
You can also make modifications from the command line using a text manipulation language like sed. For example if in file2 chr4 is represented as chrIV, than you could fix like so:
sed 's/chrIV/chr4/g' file2
http://www.grymoire.com/Unix/Sed.html
Hi
Thank you both for the responses. The data I'm using is previously published data generated using SNP chips (largely illumina). I've been advised by a colleague to check each dataset against the rsid for the chip it was generated on (to make sure nothing strange is happening), then update the ids, positions and chr using plink against the h37 and finally merge all the files right at the end.