Error occurence after merging files with bcftools: wrong number of fields ?

Question

[Lukáš]

I have multiple vcf of CASES and CONTROLS variations annotated by VEP, SNPEff, SnpSift.

first pair vcf -> only variations| CASES and CONTROLS
second pair vcf -> variations + SnpEff | CASES and CONTROLS
third pair vcf-> variations + SnpEff+VEP+SnpSIFT| CASES and CONTROLS | at so on

Because it have non-overlapping samples, i tried to merge corresponding vcf of CASE and CONTROLS into one file with bcftools merge.

bgzip -c file.vcf > file.vcf.gz
tabix -p vcf file.vcf.gz
bcftools merge -o merged.vcf.gz -Oz file.vcf.gz file_2.vcf.gz

Unfortunatelly in files with SnpSift annotations i got this error:

Error at chr1:20724637: wrong number of fields in dbNSFP_ada_score?

Please anybody knows how to solved this error or what does it mean?

[Pierre Lindenbaum]

what is the output of the following cmds:

bcftools view --header-only  file.vcf.gz | grep dbNSFP_ada_score
bcftools view --header-only  file_2.vcf.gz | grep dbNSFP_ada_score
bcftools view -G  file.vcf.gz "chr1:20724637"
bcftools view -G file_2.vcf.gz "chr1:20724637"

[Lukáš]

grep header:

INFO=<ID=dbNSFP_ada_score,Number=A,Type=String,Description="Field 'ada_score' from dbNSFP">

INFO=<ID=dbNSFP_ada_score,Number=A,Type=String,Description="Field 'ada_score' from dbNSFP">

-G on chr1:

CHROM POS ID REF ALT QUAL FILTER INFO

chr1 20724637 rs1280663306;rs1445195248 TGGGAGGGAGGGAGAGAGGT TGGGAGGGAGAGAGGT,TGGGAGGGAGGGAGGGAGGT,TGGGAGGGAGGGAGAGAGGG,TGGGGGGGAGGGAGAGAGGT,TGGGTGGGAGGGAGAGAGGT,TGGGAGAGAGGGAGAGAGGT 2.42927 . ANN=TGGGGGGGAGGGAGAGAGGT|splice_region_variant&intron_variant|LOW|SH2D5|SH2D5

...SNV|1||||||||||||||||||||||||||||||||||||||||||||;dbNSFP_ada_score=1.43020658327008E-4,0.00157457299236331,5.16496450143413E-5;dbNSFP_rf_score=0.016,0.054,0.0;dbNSFP_PHRED=5.632,0.077,6.365,7.459,2.716

CHROM POS ID REF ALT QUAL FILTER INFO

chr1 20724637 rs1280663306;rs1445195248 TGGGAGGGAGGGAGAGAGGT TGGGAGGGAGAGAGGT,TGGGAGGGAGGGAGGGAGGT,TGGGAGGGAGGGAGAGAGGG,TGGGGGGGAGGGAGAGAGGT,TGGGTGGGAGGGAGAGAGGT,TGGGAGAGAGGGAGAGAGGT 2.42927 . ANN=TGGGGGGGAGGGAGAGAGGT|splice_region_variant&intron_variant|LOW|SH2D5|SH2D5|transcript|NM_001103161.2|protein_coding|5/9|..

..||||||||||||||||||||||||||;dbNSFP_ada_score=1.43020658327008E-4,0.00157457299236331,5.16496450143413E-5;dbNSFP_rf_score=0.016,0.054,0.0;dbNSFP_PHRED=5.632,0.077,6.365,7.459,2.716

I had to cut it because it wouldnt be readeable and i think it would be too long for response. I ll try do add it into question.

[Lukáš]

Unfortunately content is over limit. Still only one differrence i have there dbNSFP annotations from others is separated by ; instead |

My last idea is that it could be only some difference beetween CASE and CONTROLS files.

Answer 1

[Pierre Lindenbaum]

your vcf is badly annotated.

there is a VCF with 6 ALT alleles :

chr1 20724637 rs1280663306;rs1445195248 TGGGAGGGAGGGAGAGAGGT TGGGAGGGAGAGAGGT,TGGGAGGGAGGGAGGGAGGT,TGGGAGGGAGGGAGAGAGGG,TGGGGGGGAGGGAGAGAGGT,TGGGTGGGAGGGAGAGAGGT,TGGGAGAGAGGGAGAGAGGT

and the annotation for dbNSFP_ada_score is INFO=<ID=dbNSFP_ada_score,Number=A,Ty..:

so Number=A implies that there MUST be one value for each ALT allele.

but there is only 3 values for this variant: 0.016,0.054,0.0 while there is 6 ALT alleles.

[Lukáš]

Thank you very much. So it is badly annotated. But it only goes for dbNSFP scores anotated by SnpSift. So do i have to change whole annotation of vcf file, or only the dbNSFP part of it?

I am sorry if its trivial but i am only familiar with annotation concept.

Once again thank you so much for your answer

[Pierre Lindenbaum]

I would remove the annotation of the two vcf files (bcftools annotat -x '...'), merge the files and re-annotate