Reformat minimac imputated files to plink file

Question

[Floris Brenk]

Hello all,

I am imputating my genotyped dataset, genotyped with illumina hapmap610 chip with minimac and mach according to standard protocols

I was wondering if anyone knows how to convert the output files of this imputation packages into standard plink format (ped and map files) with as side note that some filtering can be performed on the probability score that minimac gives as output.

Thanks in advance!

[xujiayuan32257406]

I have met the same question !! Convert the .dos, .info formate of mimimac to the plink binary formate !! Have you solved that ??

Answer 1

[Floris Brenk]

Found it: http://www.complextraitgenomics.com/software/gcta/data_management.html

gcta64 --dosage-mach test.mldose.gz test.mlinfo.gz --make-bed --out test

[xujiayuan32257406]

I have tried gcta64, but the generated .bim formate didn't have the rsID, only the chr:bp position, still can't the following analysis? Do you know how to convert to the binary .bim formate?

[Xu1912]

Is it because the dose file does not have rsid? There is an option --rsid in Minimac3, which will use rsid instead of chr:bp in the output files including dose and info file.

Answer 2

[Maxime Lamontagne]

I think dose2geno is the software you need.

Answer 3

[oyhel]

I am wondering if you managed to figure out how to filter only high quality imputed genotypes from minimac3 output. I am struggling with the same problem you did. The dose2geno does provide best guess genotypes, but it does not seem to consider low quality scores. Using PLINKs vcf-min-gp flag is not working since plink expects 0-1 scale GP but minimac3 outputs phred scale. Any help would be highly appreciated if you managed to sort this!

[xujiayuan32257406]

gcta --imput-rsq 0.3 is OK

[Floris Brenk]

Hmm cant really remember how sorry, but I did filtered out all snps with a r2 score of lower than 0.3

Answer 4

[Floris Brenk]

Yes this looks indeed interesting. Although still a bit of reformatting needs to be done to get the actual .ped and .map files... One thing that still misses is that the probability score of the minimac output is not used now and I dont know what dose2geno does with imputed low probability scored genotypes...

Answer 5

[ISB]

Use fcgene to convert minimac output to PED/MAP files. You can also use rsq as filter:

fcgene --minimac-prob minimac_output.prob --minimac-infominimac_output.info --oformat plink --rsq 0.3 --out plink_files

[xliu4255]

Thanks but fcgene requires pedinfo.txt and snpinfo.txt as input files. How can I prepare these 2 files using Minimac3 output files? And Minimac3 seems to not output .prob file.