Usage

SV² is designed for human whole genome short-read sequencing libraries. Given deletion and duplication positions, SV² returns a VCF with predicted copy number genotypes.
Whole genome alignments from the 1000 Genomes Project were used for training. Validated genotypes were obtained from the phase 3 integrated structural variation call set (DOI:10.1038/nature15394; PMID: 26432246).
Features for genotyping include coverage, discordant paired-ends, split-reads, and heterozygous allele depth ratio.
- BAM files must have supplementary alignment tags (SA).
- SNV VCF must contain Allele Depth (AD/DPR). SV² can accommodate GATK Haplotype Caller and FreeBayes VCFs.
  - SNV VCF must be compressed and indexed with bgzip and tabix
SV² operates with a bi-allelic model with a copy number range of 0-4
Output is in VCF format.
- Median Phred-adjusted ALT likelihoods are reported in the QUAL column
- SV² standard filters are reported in the FILTER column
- SV² stringent filters for de novo discovery are located in the INFO column as DENOVO_FILTER=
- Positions are annotated based on their overlap to genes, RepeatMasker, segmental duplications, 1000 Genomes phase 3 CNV, and more
SVs with estimated autosome copy number >10 cannot be genotyped.

Provide feedback