conform-gt error "Non-unique marker ID: ."

Question

[cg_ref_database]

I am using conform-gt from Beagle5 to align my input files to the reference. Conform-gt has successfully aligned the target chromosome files 1-23 to the reference using the same command below (with the exception of changing the chromosome number to match the reference, of course), but it has failed for chr2.

The two positions listed in the error below are not present in the target file, but are present in the reference file.

Does anyone have any ideas on why this error is occurring? Note: The reference file was downloaded from Beagle's website and has not been altered in anyway.

Command entered:

conform-gt.24May16.cee.jar ref=chr2.1kg.phase3.v5a.vcf.gz gt=chr2.vcf.gz chrom=2 out=mod.chr2 excludesamples=non.eur.excl

Error:

Exception in thread "main" java.lang.IllegalArgumentException: Non-unique marker ID: .
2       11656437        .;rs531258135   GGTGTGTGTGTGTGT GGTGTGTGTGTCTGT,G
2       18004148        .;rs541466601   AGAGCCCA        AGAGCCCG,A
        at conform.ConformMarkers.idMap(ConformMarkers.java:135)
        at conform.MatchedMarkers.<init>(MatchedMarkers.java:66)
        at conform.ConformMain.<init>(ConformMain.java:114)
        at conform.ConformMain.main(ConformMain.java:97)

Here are a few lines of the target vcf (there are over 100 samples in the target file, but only three are shown):

2       1146828 rs4971393       C       A       .       .       PR      GT      0/0     0/0     0/0
2       1154664 rs4467306       A       G       .       .       PR      GT      0/1     0/0     0/1
2       1287872 rs13411595      A       G       .       .       PR      GT      0/1     0/0     0/0
2       1321795 rs13033766      A       G       .       .       PR      GT      0/1     0/1     0/1
2       1392629 rs4261760       G       A       .       .       PR      GT      0/0     0/0     0/0 

For the chromosome files, chr1, chr3-chr23, that conform-gt did successfully align to the reference genome, the

[4galaxy77]

Seems like you possibly have some weird IDs. Can you paste into your question a couple of lines of your target vcf (without the header).

[cg_ref_database]

I have added some lines of the target file to my question.

Answer 1

[Jack]

I had this issue too, on the same chromosome. I found out that the snv 11656437 on chr2 was multiallelic and so appeared in several records in that reference panel (beagle and conform-gt don't allow this). The solution I used was: before using conform-gt you need to make sure each reference panel contains only biallelic snvs. You can use bcftools for this:

$ bcftools view -m2 -M2 -v snps -Oz -o chr2.refPanel.biallelic.vcf.gz chr2.refPanel.vcf.gz 

and then use chr2.refPanel.biallelic.vcf.gz as the reference panel input to conform-gt, rather than chr2.refPanel.vcf.gz.

Hope this helps!

[cg_ref_database]

Thank you, Jack! Your suggestion worked.

[Jack]

glad I could help :)