vt partition reporting no overlap

[cboustred]

Hi,

I'd like to use vt partition to identify the similarities and differences between variants called by different variant callers on the same set of samples.

However I'm having some issues...

• using VT to compare platypus and freebayes vt partition platypus_chr22_normal_filt.vcf.gz freebayes_chr22_normal_filt.vcf.gz
• returns this:

partition v0.5 Options: input VCF file a platypus_chr22_normal_filt.vcf.gz input VCF file b freebayes_chr22_normal_filt.vcf.gz A: 663 variants B: 609 variants ts/tv ins/del A-B 663 [3.04] [1.20] A&B 0 [-nan] [-nan] B-A 609 [3.12] [1.50] of A 0.0% of B 0.0% Time elapsed: 0.02s

i.e. there are zero overlaps between variants in the two files

However, using vcftools I can see there are ~565 variant overlaps between variants in the two files:

vcf-isec -f platypus_chr22_normal_filt.vcf.gz freebayes_chr22_normal_filt.vcf.gz | grep -v ^# | wc -l 565

Any thoughts as to why vt partition would not be working? It appears to be an issue with the Freebayes VCF as when I do a comparison with a VCF generated with VarScan2 I get results as expected from vt parition:

vt partition varscan_chr22_normal.vcf platypus_chr22_normal_filt.vcf.gz

partition v0.5 Options: input VCF file a varscan_chr22_normal.vcf input VCF file b platypus_chr22_normal_filt.vcf.gz A: 673 variants B: 663 variants ts/tv ins/del A-B 97 [2.36] [1.00] A&B 576 [2.93] [1.50] B-A 87 [3.73] [0.75] of A 85.6% of B 86.9% Time elapsed: 0.02s

Using vt partition + varscan vcf + freebayes vcf returns 0 overlap hence why I think it is an issue with the freebayes VCF.

I have run vt partition comparing the freebayes vcf to itself and that returns as expected results: vt partition freebayes_chr22_normal_filt.vcf.gz freebayes_chr22_normal_filt.vcf.gz partition v0.5

Options: input VCF file a freebayes_chr22_normal_filt.vcf.gz input VCF file b freebayes_chr22_normal_filt.vcf.gz A: 609 variants B: 609 variants ts/tv ins/del A-B 0 [-nan] [-nan] A&B 609 [3.12] [1.50] B-A 0 [-nan] [-nan] of A 100.0% of B 100.0% Time elapsed: 0.03s

Commands for generating the Freebayes VCF were this:

• call variants freebayes --genotype-qualities -f $ref --region chr22 -L $bam_list > $out_dir/test_out_freebayes_chr22.vcf
• limit to cds regions & sort vt view -I test.intervals test_out_freebayes_chr22.vcf.gz -o tmp_freebayes_chr22.vcf vt sort tmp_freebayes_chr22.vcf -o freebayes_chr22.vcf
• normalise vt normalize freebayes_chr22.vcf -r $ref -o freebayes_chr22_normal.vcf
• filter vt view freebayes_chr22_normal.vcf -f "INFO.DP>=30&&QUAL>=20" -o freebayes_chr22_normal_filt.vcf bgzip freebayes_chr22_normal_filt.vcf tabix -p vcf freebayes_chr22_normal_filt.vcf.gz

Any help with this much appreciated

BW

Chris

[cboustred]

bcftools isec platypus_chr22_normal_filt.vcf.gz freebayes_chr22_normal_filt.vcf.gz -p dir

Also works as expected reporting ~561 shared variants using the freebayes vcf

[atks]

The contigs in the headers of the vcf files have to be the same for vt partition to work. Are the headers the same?

[cboustred]

Bingo, that did the trick! Thanks very much! Chris

[atks]

Thanks for reporting this, I'll add a check for that later.

[atks]

fixed.