bcftools merge; retaining sample names

Question

[Lee Katz]

When I do bcftools merge, the headers do not retain the filenames. How can I specify filenames?

This is my command

bcftools merge vcf/unfiltered/*.vcf.gz -O z > msa/pooled.vcf.gz

However this is the relevant part of my header, despite the filenames I gave it. Is it just up to me to parse the mergeCommand line? Or is there a way to use bcftools query to get the right headers after the fact?

##bcftools_mergeVersion=0.2.0-rc7-47-g02a1fb3+htslib-0.2.0-rc7-36-g6e2ebc4
##bcftools_mergeCommand=merge -O z vcf/unfiltered/lambda_virus.fasta.wgsim.fastq.gz-lambda_virus.vcf.gz vcf/unfiltered/lambda_virus.fasta.wgsim.fastq.gz-reference.vcf.gz vcf/unfiltered/sample1.fastq.gz-lambda_virus.vcf.gz vcf/unfiltered/sample1.fastq.gz-reference.vcf.gz vcf/unfiltered/sample2.fastq.gz-lambda_virus.vcf.gz vcf/unfiltered/sample2.fastq.gz-reference.vcf.gz vcf/unfiltered/sample3.fastq.gz-lambda_virus.vcf.gz vcf/unfiltered/sample3.fastq.gz-reference.vcf.gz vcf/unfiltered/sample4.fastq.gz-lambda_virus.vcf.gz vcf/unfiltered/sample4.fastq.gz-reference.vcf.gz 
#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  Sample1 2:Sample1       3:Sample1       4:Sample1       5:Sample1       6:Sample1       7:Sample1       8:Sample1       9:Sample1       10:Sample1

Answer 1

[Pierre Lindenbaum]

how about renaming the samples in all your input *.vcf before calling bcftools ?

[Lee Katz]

How do you do that? Just change the header from

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  Sample1

to

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT lambda_virus.fasta

?

Or is there a Vcf.pm method? Or bcftools/vcftools method?

[Pierre Lindenbaum]

sed '/^#CHROM/s/Sample1/lambda_virus.fasta/' in.vcf > out.vcf

[Lee Katz]

I can't figure out your sed magic but it essentially works, thanks! This is my full system call.

varscan.sh mpileup2cns $pileup --min-coverage $$settings{coverage} --min-coverage 10 --min-var-freq 0.75 --output-vcf 1 |\
    perl -lane 's/Sample1/\Q$vcf\E/; print;' |\
    bgzip -c > $vcf

[Ram]

Read the sed command so:

From the file in.vcf

In lines that begin with "#CHROM" (/^#CHROM)

substitute "Sample1" with "lambda_virus.fasta" (/s/Sample1/lambda_virus.fasta)

and write the output to "out.vcf" (>out.vcf)

Put together,

sed '/^#CHROM/s/Sample1/lambda_virus.fasta/' in.vcf > out.vcf

[Michal Nevo]

my command:

sed '/^#CHROM/s/unknown//storage/users/IsanaRNA/FISH_DATA/MappingToAcipenserRuthenusGenome/results/Bam_sorted/F1.sorted.bam/' five_contigs_cp.vcf > out.vcf

I get :

sed: -e expression #1, char 21: unknown option to `s'

[Ram]

You may need to use the -E option. Plus, if you have / characters in your expressions, don't use / as the sed-separator. Use something like | instead:

sed '|^#CHROM|s|...'

Answer 2

[Lee Katz]

Answer was essentially from Pierre: find and replace Sample1 with the correct name in each corresponding vcf file

[Pierre Lindenbaum]

I moved my comment to an answer