Missing homozygous deletion

[seedgeorge]

Hi,
We find Battenberg a good and sensitive tool for the detection of alterations - however we've run into a possible edge case where a homdel (at the start of chromosome 9) is missed due to having no snps (at all) passing the depth filter (set to 30) in this area. I'm attaching a figure which I hope will be informative. This is a high-purity case so may not be representative of many tumour samples - but is real nonetheless.

We've explored modifying the depth filters here but really it'd be best to solve this directly, although I appreciate the challenge in this case as lacking allele-frequencies may limit downstream steps.

Can provide more files/details if that would help.

Kind regards,
George

[seedgeorge]

I also have noticed that as Battenberg doesn't segment using log2 ratios, only the allele phased allele frequencies (unless I'm missing something), there is a high chance of missing homozygous deletions that are adjacent or within copy-neutral segments - because there isn't an allele frequency shift for the algorithm to segment.

I attach a second plot with the missed deletion highlighted.

Perhaps this issue would be resolved by switching from multipcf to asmultipcf?