Removing multi-variant records from vcf file

Question

[Emili]

I am using gatk ASEReadCounter to get the read counts per allele. To do so, I used the following command:

gatk ASEReadCounter  -R /path_to_genome/hg38_genome/GRCh38.p13.genome.fa  -I sample.sorted.bam
-V sample.vcf.gz  -O output.table

I used GATK4. but I realized In my VCF at position chr1:1574033, there are more than one variant record in the VCF. This is not accepted by ASEReadCounter. if it was only one, I could do it manually but the question is how can I remove those rows if many rows have more than one variant record. does GATK have such ability? the row in my vcf file looks like this:

chr1    1574033 .   AAG *,A 55.01   .   AC=1,1;AF=0.500,0.500;AN=2;DP=12;ExcessHet=3.0103;FS=0.000;MLEAC=1,1;MLEAF=0.500,0.500;MQ=60.00;QD=6.88;SOR=1.179   GT:AD:DP:GQ:PL  1/2:0,6,2:8:29:434,65,29,147,0,111

[cpad0112]

I would suggest to normalize the VCF. This way multiallelic records will be flattened and each row would have only one allele. bcftools can do that.

Answer 1

[drabiza1]

bcftools norm -m - sample.vcf.gz > normalized_sample.vcf

Answer 2

[JustinZhang]

If you are handling with germline variants, pre.py in (github) Illumina/hap.py can help you normlize your vcf file almost perfectly.

If you are handling with somatic variants, try bcftools or other software. (not verified)

Answer 3

[geocarvalho]

Another option is GATK SelectVariants and check the select-type-to-include and restrict-alleles-to parameters :

docker run -v $PWD:$PWD -w $PWD broadinstitute/gatk:4.4.0.0 gatk SelectVariants \
     -R GRCh38.genome.fa \
     -V ${SAMPLE}.vcf.gz \
     --restrict-alleles-to BIALLELIC \
     --select-type-to-include SNP \
     -O ${SAMPLE}.selectvariants.vcf.gz