Error: kmermatcher step died

[gbdias]

Hi @martin-steinegger, thanks for making conterminator available.

I've been trying to run it on a 450 GB fasta file, using 20 CPUs and 450 GB of RAM, but keep getting the following error:

./15038853176956681317/conterminatordna.sh: line 47: 64706 Killed                  $RUNNER "$MMSEQS" kmermatcher "$TMP_PATH/db_rev_split" "$TMP_PATH/pref" ${KMERMATCHER_PAR}
Error: kmermatcher step died

• I suspect this error comes from high memory usage since the peak for this job was 469,875,968kb for memory and 1,185,770,492kb for virtual memory. Do you agree with this?
• Could I remediate this situation by using fewer parallel threads or the only way is to provide more memory?

Thanks

[glucksfall]

Similarly to @gbdias, I found that using conterminator with a nucleotide fasta also fail:

Segmentation fault (core dumped)                                  ] 0.00% 1 eta -
Error: crosstaxonfilterorf step died

The error happens with both pre-compiled versions of conterminator and with a locally compiled version of the software. I tested the software on a single dna fasta (also tested with a protein fasta derived with prodigal, failing also in the kmermatcher step).

That fasta is an assembly from SPAdes that I suspect is contaminated with a second species. I would like to corroborate and include conterminator in my default analyses. Hope we could understand better the source of the malfunction and correct it.

Thanks @martin-steinegger, best regards.