Variant Calling Issues

[FilippoLucchini]

Good morning,

We recently tested LinkedSV for SVs calling in human WGS samples (data comes from TELL-Seq Linked-Reads libraries). When looking at calls, we noticed three possible issues:

• one 32kb homozygous deletion was not called, while it's reported in the BAM file (i attach a picture).
• No translocations or inversions were called (we know there is at least one known translocation). We saw there's already an open issue about it, we'll check for the dbo score in temporary files and see.
• Speaking about duplications, we got some calls in the large.cnv.bedpe file, but a known tandem duplication of 13kb wasn't called (while it's present in the bam file, see the second picture).

Due to the fact that LinkedSV input is a phased.sorted.bam, does it call only variants which are correctly phased?
Or may be related to variant filtering steps?
Looks strange that it only called 1-3 large structural variants for each sample (in one sample, no large sv where called at all)

Please note that i'm not a bioinformatician, my collegue is on holiday at the moment and i'm opening this issue for him.
He told me that LinkedSV was launched with default parameters

Thanks in advance

Kind Regards

.