GATK SelectVariants consider spanning/overlapping deletions as SNP

Question

[cocchi.e89]

Quick question, I splitted a multiallelic VCF file with bcftools:

bcftools norm -m -any <IN.vcf> -OV > <OUT.vcf>

and then divided SNP from INDEL with GATK SelectVariants:

gatk SelectVariants \
 -R <REFERENCE.fasta> \
 -V <OUT.vcf> \
 --select-type-to-include SNP \
 -O <OUT.SNP.vcf>

But I noticed that this SNP-only VCF includes spanning/overlapping deletions (* allele) as SNP. As example:

chr1    10443   .   C   *   54.40   VQSRTrancheSNP99.90to100.00 AC=1;AF=0.038;AN=26;ANN=T|upstream_gene_variant|MODIFIER|DDX11L1|ENSG00000223972|Transcript|ENST00000450305|transcribed_unprocessed_pseudogene|||||||||||1567|1||sequence_alteration|HGNC|HGNC:37102||||chr1:g.10443C>T,T|upstream_gene_variant|MODIFIER|DDX11L1|ENSG00000223972|Transcript|ENST00000456328|processed_transcript|||||||||||1426|1||sequence_alteration|HGNC|HGNC:37102|YES|||chr1:g.10443C>T,T|downstream_gene_variant|MODIFIER|WASH7P|ENSG00000227232|Transcript|ENST00000488147|unprocessed_pseudogene|||||||||||3961|-1||sequence_alteration|HGNC|HGNC:38034|YES|||chr1:g.10443C>T;BaseQRankSum=1.83;DP=336;ExcessHet=0.202;FS=3.31;InbreedingCoeff=0.4448;MLEAC=2;MLEAF=0.077;MQ=30.71;MQRankSum=0;PG=0,8,19;QD=9.07;ReadPosRankSum=0.842;SOR=0.105;VQSLOD=-7.763;culprit=MQ    GT:AD:DP:FT:GQ:PL:PP    0/0:27,0:27:PASS:32:0,24,360:0,32,379   0/0:20,0:20:lowGQ:8:0,0,161:0,8,180 0/0:29,0:29:lowGQ:8:0,0,654:0,8,673 0/0:24,0:24:lowGQ:8:0,0,458:0,8,477 0/0:12,0:12:PASS:35:0,27,405:0,35,424   1/0:2,2:6:PASS:55:136,65,63:118,55,64   0/0:22,0:22:PASS:59:0,51,765:0,59,784   0/0:43,0:43:lowGQ:8:0,0,653:0,8,672 0/0:42,0:42:lowGQ:8:0,0,810:0,8,829 0/0:32,0:32:lowGQ:8:0,0,410:0,8,429 0/0:36,0:36:PASS:38:0,30,846:0,38,865   0/0:28,0:28:PASS:59:0,51,765:0,59,784   0/0:15,0:15:lowGQ:8:0,0,265:0,8,284

I think this is incorrect, aren't those supposed to be DEL? Or am I wrong?

Thank you in advance for any help!

Answer 1

[Pierre Lindenbaum]

I think this is incorrect, aren't those supposed to be DEL? Or am I wrong?

it's not an indel, it's IN an indel (!). it is a local haploid region with a variant (you removed the ALT allele with norm) but there should a variant with a large deletion upstream of "chr1 10443"

[cocchi.e89]

Thanks so much. So can I consider a haploid region as SNP?

[Pierre Lindenbaum]

well, that variant chr1 10443 . C * is meaningless without the associated ALT. It should be discarded.

[jon.klonowski]

If norm is removing the orginal indel, doesn't it suggest that the variant remaining after norm is a 1bp deletion, but that now it is misnomered with a C -> * ? Trying to understand how would a indel be deleted by norm

[jon.klonowski]

I am having a similar issue, where I find variants from a phased trio VCF being denoted by * but without an upstream deletion on that same phased allele. I want to get rid of these SNPs but not sure how.

Here is an example. In the trio VCF - so 2 parents and the patient - I find:

#CHROM  POS      REF     ALT
chr1    154590147  CCG     C
chr1    154590148  CG      C
chr1    154590149  G       *
chr1    154590149  G       C

and then, when I just extract the patient genotypes using bcftools query:

#CHROM  POS         REF   ALT     GT
chr1     154590148   CG  C      0|1
chr1     154590149   G   *      1|0
chr1     154590149   G   C      0|1

see my question: Removing / Excluding / Collapsing Overlapping Indels