After incorporating reference selection changes in #528, the workshop exercise of LASV assembly now fails for two samples that it did not previously fail on. This is because the example data against the supplied reference genomes is too far for skani to hit, and the resulting skani dist output is empty, resulting in CHOSEN_REF_FASTA being unset.
The reference selection code in task assembly needs to have a sensible fallback behavior (perhaps the pre-#528 behavior or maybe something mummer or minimap2 based) in the scenario when the skani dist output is empty. The error mode shouldn't happen, and even if it did, the error messages are cryptic.
dpark01
commented
3 weeks ago
After incorporating reference selection changes in #528, the workshop exercise of LASV assembly now fails for two samples that it did not previously fail on. This is because the example data against the supplied reference genomes is too far for skani to hit, and the resulting
skani distoutput is empty, resulting inCHOSEN_REF_FASTAbeing unset.The reference selection code in
task assemblyneeds to have a sensible fallback behavior (perhaps the pre-#528 behavior or maybe something mummer or minimap2 based) in the scenario when theskani distoutput is empty. The error mode shouldn't happen, and even if it did, the error messages are cryptic.