Possible bug in bin sizes

[gevro]

Hi,
I ran the latest version of cnvkit via docker with these settings:

cnvkit.py batch -n -f REF.fasta -m hybrid -g ACCESSIBLEREGIONS.bed -t PROBESBED.bed --target-avg-size 50000 --antitarget-avg-size 4000000 --antitarget-min-size 1000000

However, it is producing a target.bed file whose average bin size is only 122.79 bp. This should not be possible based on the above settings.

This is with hg38.

So there appears to be a bug in the way batch interprets these parameters:
--target-avg-size 50000

Or is this because bins cannot encompass more than one target? If so, I assume then that the segment step integrates information across multiple bins?

Thanks

[etal]

You got it, the target bin boundaries are limited by the regions in your PROBESBED.bed. If your panel captures exons with an average size of 122.79 bp, that will be the average in the output. The --target-avg-size parameter works the way you would think in WGS and in large captured regions like introns (e.g. if your panel is looking for fusion sites, not just the exome/exons).

Segmentation is where the signal is extracted across multiple adjacent targets and antitargets, so segments can be up to the length of a whole chromosome.