search

llecompte / SVJedi

SV genotyping with long reads
GNU Affero General Public License v3.0
40 stars 4 forks source link

Error when generating allele sequences from a VCF obtained with GRIDSS #11

Closed clemaitre closed 2 years ago

clemaitre commented 2 years ago

A bug reported by a user, with an input VCF file output by the SVcaller GRIDSS.

Some BND variants in the VCF file are not well recognized by SVJedi and causes the following error: 

File "modules/generateRef.py", line 333, in translocation_cases
    return case
UnboundLocalError: local variable 'case' referenced before assignment
clemaitre commented 2 years ago

Note: GRIDSS output VCF files are not well formated for SVJedi. SV types are not annotated: each SV is represented by one or several BND lines in the vcf file.

For SV types containing multiple BND lines, as is the case for example with a deletion or an inversion, this is not suitable for SVJedi, as each vcf line is considered as a different SV and is treated independently of the others. In this case, if the size of the SV is smaller than the -ladj parameter (default = 5 Kb), this may lead to the incapacity to genotype the variant (or the 2 BND lines), as the reads mapping on the 2 variants will be considered as multi-mapped.