samuell
commented
3 months ago Based on some testing with a separate workflow that tries to mimic the steps of microSALT for assembly and resistance blasting, it turns out the reason VIM is not reported is because of too small alignment lenghts in the blasting against the beta-lactamase file from ResFinder db. Because of the thresholds for the report (97% identity and 90% alignment length), the matches therefore don't end up in the report, although some matches are found in the blast results.
I did test a lot of combinations of assemblers (Spades vs Skesa) and options such as (Trimming vs No trimming) and the various assembly mode flags to Spades. I will exclude other flags than --isolate here though, since we need to use that to fix the arcC issue with NovaSeq X, but I found this:
- With Spades with the
--isolateflag, the alignment length increased a bit for VIM genes when turning off trimming (from ca 433bp to ca 612bp), although it is not above 90% of the full VIM length at 801bp for one of the variants, which is our treshold. - With Skesa, the alignment lengths instead improved when including trimming. From ca 591bp without trimming, to 801bp (100%) when including the trimmomatic trimming with the parameters used in microSALT.
Thus, the results suggest that switching the assembler from Spades to Skesa and keeping the trimming active might potentially resolve this issue.
Describe the bug
For some datasets where ResFinder ran on the raw reads report blaVIM beta lactamase resistance genes, there is no report of VIM genes at all in microSALT.
KS reports:
To Reproduce Steps to reproduce the behavior:
24ET500149with microSALT 3.3.5Expected behavior The report for "24ET500149" should include
blaVIM-4.Software version (please complete the following information):
Additional context