Hello,
I was wondering if there is anybody to help me filter my vcf file (from freebayes) in order to check for heterogeneity in my single genome sequencing data.
I assume I should use:
vcftools --vcf my.vcf --maf 0.4 --recode --recode-INFO-all --out my_filtered.vcf
but I am not sure how I can add an option for low coverage reads.
Thanks!
•
written
by
pennakiza
1 answer
Since it looks like you are using a single-sample VCF, this is pretty straightforward. Just add something like --min-meanDP n, where n is the depth cutoff, to your vcftools command to filter out low depth sites. From the vcftools documentation:
--min-meanDP <float>
--max-meanDP <float>Includes only sites with mean depth values (over all included individuals) greater than or equal to the "--min-meanDP" value and less than or equal to the "--max-meanDP" value. One of these options may be used without the other. These options require that the "DP" FORMAT tag is included for each site.
Log in to answer this question.
-
Bcftools: filtering by AD as a percentage of DP
written by Timotheus •
-
unable to impute with beagle after filtering VCF table using vcftools
written by ziv_attia •
-
Why is VCFtools excluding invariable sites?
written by raf.marcondes •
-
Different estimates of nucleotide diversity (pi) from two pipelines: pixy vs vcftools
written by nitinra •
-
Filter max-missing multiple time removes more SNPs
written by QPaps04 •
-
Warning: Expected at least 2 parts in INFO entry...while using vcftools, what could be the reason
written by nidhisukhija5 •
-
VCFTOOLS error while trying to filter for MAF
written by lafzi.a •
-
Filter SNPs on depth
written by Covux •
-
vcf SNP file minor allele frequency filter?
written by sun.nation
-
Convert VCFf file to plink ped/map format while filtering the SNPs
written by roll
Can you provide more detail on your starting VCF, and exactly how you would like to filter it?
Depth of coverage is (generally) a sample-level statistic. There isn't one coverage value for the whole variant row; there's a value for every genotype call in the row. Assuming you have a multi-sample VCF, what do you want to do with low coverage genotypes? Remove the whole row if any genotypes have low coverage? Change the genotype to missing if the coverage is low? Etc.
On the other hand, you may have a DP statistic in the INFO field that represents something like the mean coverage across all samples; is this what you're referring to?
So, I've got a vcf file of a single genome (HIV provirus) and what I would like to do is to confirm that there is only one genome indeed, by checking if there are any accurate variants. In this context, I thought that if I get rid of any rows of low coverage variants and minor alleles that appear less than 40%, I can get an idea of how "clean" my sample is. My VCF looks like this: