Question: Clarification on building a reference about cnvkit HOT 3 CLOSED

For WGS on the post-BQSR BAM where all reads in difficult/repetitive regions have been removed, you may need to run in "amplicon" mode with the accessible regions as your "targets" in order to force CNVkit to skip those regions. Otherwise, I think "wgs" mode ignores sequencing accessibility since that was originally intended to be a filter on off-target regions that were not considered in the capture design. Or, you can run it on the pre-BQSR BAM in "wgs"; the main difference is the default bin size.

For WES or panel, yes, you can run batch with the capture targets BED as input and it will figure out appropriate bin sizes and sequencing accessibility. I recommend using the pre-BQSR BAM as input for these samples because CNVkit uses the copy number signal from off-target reads, and handles "difficult" genomic regions in its own way. If your pipeline has already filtered out all reads in the the off-target and difficult regions, then these post-BQSR BAMs would be better analyzed in amplicon mode with the intersection of capture targets and accessible regions as the input -- i.e. all the locations where there are still reads remaining after the filter.

You don't strictly need to use --male-reference (a.k.a. haploid X reference) to analyze male samples. The calls and exported VCF will work the same either way (assuming the option is used consistently throughout the pipeline); the main difference is that the plots will show haploid X as neutral copy number, e.g. scatter will plot a single copy of X with a log2 value of 0.0 instead of -1.0. It's kind of a legacy feature.

from cnvkit.

Awesome, thank you so much for the detailed answer 🙏 😃

from cnvkit.