Eror combining vcf files

[assafgrw]

Hello,

I am trying to generate Mendelian error statistics for trio analysis i do. According to Seqmule guide I should first combine the vcf of the mother father and son. So I figured it would make most sense to combine the consensus vcf of each.
When using the command "./seqmule stats --u-vcf father.vcf, mother.vcf, son.vcf -p 123combo -ref hg19.fa"
I got an error which says that the reference does not exist so I used human_g1k_v37_fasta as reference:
"./seqmule stats --u-vcf father.vcf, mother.vcf, son.vcf -p 123combo -ref human_g1k_v37.fasta" but get the following eror:

"ERROR: not all VCFs have the same set of samples (case-sensitive)!"

Any suggestions?
Kind regards
Assas

[assafgrw]

I was also trying to run it now only on the freebayes vcf's and got the same error

[yunfeiguo]

@assafgrw thanks for reporting the issue, I will address it soon.

[assafgrw]

thanks

[yusmile0618]

I also meet this error.
ERROR: not all VCFs have the same set of samples (case-sensitive)!
Is this error fixed?
After using GATK combinevariants. I can got the mendel_stat results, but I don't know the threshold of Proportion of Mendelian errors. Where can I find these infomation ?
thanks.

[yunfeiguo]

Hi @yusmile0618,

are you running with --u-vcf or --c-vcf? which VCFs are you using? these 2 options are for merging VCFs of the same sample.

Mendelian errors, in ideal scenario, should be zero. So the lower it is the better. Thanks.

[yusmile0618]

thanks @yunfeiguo
seqmule stats --u-vcf son.extract_consensus.vcf,mother.extract_consensus.vcf,father.extract_consensus.vcf -p fam -ref pathtoSeqMule/database/human_g1k_v37.fasta
both --u-vcf, --c-vcf, -u-vcf, -c-vcf get the ERROR: not all VCFs have the same set of samples (case-sensitive)!

the vcfs files are all generated from the command
seqmule pipeline -a R1.fq.gz -b R2.fq.gz -e -q -t 5 -prefix son -capture pathtoSeqMule/database/hg19agilent/hg19_SureSelect_Human_All_Exon_V6_r2.bed.

Q1: how to merging sample from 3 or more different samples?
Q2: for the mendel_stat.txt file,there are three Mendelian errors, father and offspring, mother and family, which error rate threshold can be used ? 0.01 or 0.05 ?

[yunfeiguo]

Hi merging VCFs from different samples is not supported by seqmule right now. the best way is use multi-sample variant calling on FASTQs from multiple samples, then each VCF will contain multiple samples. e.g.

seqmule pipeline -a fa_R1.fq.gz,mo_R1.fq.gz,son_R1.fq.gz -b fa_R2.fq.gz,mo_R2.fq.gz,son_R2.fq.gz -ms -e -q -t 36 -prefix father,mother,son -capture default

I am not aware of widely-used threshold for Mendelian error. However, it should be correlated with the sequencing depth, quality of variant calls of your analysis. I would expect Mendelian error is in the same order of magnitude as variant calling errors.

[yusmile0618]

@yunfeiguo ok, thanks for your multiple samples suggestion.