Converting between Impute2 and Ped/Map after imputation - 1st column dashes (--) problem

Question

[Philip Robinson]

Dear All,

have taken a PED/MAP format PLINK file and converted it into a .gen/.sample file with gtool. This has given me this look:

pkd@bioinform:~/strand_correct_script/Files_during_updating$ head controls.gen | cut -d " " -f 1-20

5 chr5:96000607 96000607 A G 1 0 0 1 0 0 1 0 0 0 1 0 1 0 0
5 rs1421911 96000947 C T 0 1 0 0 0 1 0 1 0 0 0 1 0 1 0
5 rs6860934 96001842 C T 0 1 0 0 1 0 0 1 0 1 0 0 0 0 1

I understand from the IMPUTE website the first column should be SNP1,SNP2,SNP3, but I pushed on thinking maybe things would sort themselves out. I imputed with IMPUTE2 against 1000 genomes and then this produced this format of .gen file:

pkd@bioinform:~/Impute2/converting_back_to_plink$ head European_imputed_controls.gen | cut -d " " -f 1-20

--- 5-96000097 96000097 A G 1 0 0 1 0 0 0.976 0.024 0 1 0 0 1 0 0
--- 5-96000203 96000203 C T 1 0 0 1 0 0 1 0 0 1 0 0 1 0 0
--- 5-96000264 96000264 C T 1 0 0 1 0 0 0.998 0.002 0 1 0 0 1 0 0
--- rs7733671 96000269 G A 0 1 0 0 1 0 0 0.947 0.052 1 0 0 0 0 1
--- 5-96000338 96000338 C A 1 0 0 1 0 0 0.997 0.003 0 1 0 0 1 0 0
--- rs73774358 96000463 A G 1 0 0 1 0 0 0.985 0.015 0 1 0 0 1 0 0
--- 5-96000525 96000525 G A 1 0 0 1 0 0 0.997 0.003 0 1 0 0 1 0 0
5 chr5:96000607 96000607 A G 1 0 0 1 0 0 1 0 0 0 1 0 1 0 0
--- rs73774359 96000658 A C 1 0 0 1 0 0 0.985 0.015 0 1 0 0 1 0 0

When I tried to convert it back to PLINK PED/MAP Plink said that all the SNPs were named the same "---" and crashed in flames. I have read the gtool site and cannot see any reference to what it puts in the first column when it converts from PLINK to GEN/SAMPLE, or what IMPUTE2 should do when it imputed new snps. I can load the file into R and put an arbitrary first column in, but I was wondering whether this is necessary or have I made an error somewhere.

Thank you in advance.

Philip

Answer 1

[chrchang523]

PLINK 1.9 has --recode oxford for direct export, and --data/--gen/--bgen/--sample for import. --hard-call-threshold can be used to set a genotype likelihood cutoff, or randomize genotypes based on the likelihoods, during import.

Answer 2

[Caddymob]

I have had this problem... I wrote a quick and dirty perl script to get past this. Nothing fancy, but it works. This is meant to do this by chromsome - I split these up on a cluster computer, but hopefully this gets you going.

#!/usr/bin/perl -w

$file = $ARGV[0];
$chr = $ARGV[1];

open(FILE,"<$file") || die;

while(<FILE>) {
    chomp($_);
    ($CHR,$SNP,$ZERO,$POS) = split;
    $ZERO = "0";
    if ($SNP =~ "---") {
        $SNP = "$chr:$POS";
        if (exists $snp_hash{$SNP}) {
            $snp_hash{$SNP}++;
            $SNP = $SNP . '.' . $snp_hash{$SNP};
        }
    } else {
         if (exists $snp_hash{$SNP}) {
             $snp_hash{$SNP}++;
             $newSNP = $SNP . '.' . $snp_hash{$SNP};
             $SNP = $newSNP;
         }
    }
    $snp_hash{$SNP}++;
    print "$chr\t$SNP\t$ZERO\t$POS\n";
}

Answer 3

[Maxime Lamontagne]

Do you try gtool? It can convert Impute2 output too PED/MAP format.

gtool -G --g file1 --s file2.sample --ped file3.ped --map file4.map --phenotype phenotype_1 --threshold 0.95 > output.gtool

---EDIT---

chr=1

awk -v var1=$chr '{
ORS = ""
print var1"\t"
if ($2 == "---") print "SNP."var1"."$4"\t"
else print $2"\t"
print $3"\t"
print $4"\n"
}' Chr${chr}.IMPUTE2.map > Chr${chr}.IMPUTE2.V2.map

I think this script is less complex. You can use it in a for loop with each chromosome in a separated file.

Thanks for the comment.

[Caddymob]

he did use gtool, problem is that SNPs get called "---" in the map if there is no rsID#. My script above will convert these SNPs to chr:pos format to give them a uniq ID and get you through PLINK.

Answer 4

[Kantale]

Also take a look at this python implementation:

http://www.pypedia.com/index.php/Convert_impute2_gprobs_to_PEDMAP_beagle_user_Kantale

From the long parameter list, you can only define the following parameters:

chromosome, input_impute2_gprobs_filename, input_impute2_info_filename, output_TPED_filename, output_TFAM_filename

Note: The output format is transposed PED/MAP files. You can use these files directly in plink with the --tfile parameter: http://pngu.mgh.harvard.edu/~purcell/plink/data.shtml#tr