What's the difference between --collapse and --crushvcf options of Pisces

[ShannonDaddy]

Hi, I can't tell the difference of --collapse and --crushvcf options of Pisces from the documents and command-line help, please help me out. Thanks a lot!

[tamsen]

"--collapse" will change the called variants, generally filtering out noise cause by variants broken up by read boundaries.. The idea is that variants that span multiple bases can run off the ends of reads. Ie, some reads will span the variant, so the WHOLE variant will be in the cigar string. While other reads will only overlap part of the variant, thus only partial versions of the real variant will be in the cigar string. Collapse takes these "open ended" partial variants and dumps all their read-support into the fully anchored variant. This raises the frequency of the true variant, and reduces the number of partial variants.

"-crushvcf" does not change the variants called. It just changes the formatting of the output vcf. A "crushed vcf" outputs all co-located variants on a single vcf line. An "uncrushed" vcf prints out one varaint per line, so there are frequently lines with multiple loci.

[tamsen]

Also, this might be helpful. https://github.com/Illumina/Pisces/wiki/Pisces-5.2.10-Supported-Options