Using version HUMAnN 3.7 ( tag: v3p7 ),
I built custom ChocoPhlAn database:
humann --input demo.fastq --output . --taxonomic-profile max_profile.tsv --o-log demo.log --log-level DEBUG --threads 8 --nucleotide-database /database/humann/chocophlan
The resulting database has files:
demo_bowtie2_index.1.bt2l
demo_bowtie2_index.2.bt2l
demo_bowtie2_index.3.bt2l
demo_bowtie2_index.4.bt2l
demo_bowtie2_index.rev.1.bt2l
demo_bowtie2_index.rev.2.bt2l
I run HUMANnN on a dataset using this custom database:
humann --input SAM_00015.fastq --output . --o-log SAM_00015.log --log-level DEBUG --threads 10 --nucleotide-database joint_chocophlan --bypass-nucleotide-index
and see the error:
CRITICAL ERROR: Error executing: /opt/conda/bin/bowtie2 -q -x joint_chocophlan/demo_bowtie2_index.rev -U SAM_00015.fastq -S SAM_00015_humann_temp/HOPE-SAM_00015_bowtie2_aligned.sam -p 10 --very-sensitive
Error message returned from bowtie2 :
Could not open index file joint_chocophlan/demo_bowtie2_index.rev.rev.1.bt2l
Could not open index file joint_chocophlan/demo_bowtie2_index.rev.rev.2.bt2l
I have traced the error to file, humann/humann/config.py, line 396:
bowtie2_large_index_ext=“.1.bt2l”
which is referenced in function search.nucleotide.find_index. The suffix matches more than one index file, and may result in the erroneous basename, ‘joint_chocophlan/demo_bowtie2_index.rev’. To correct this I edited to read:
bowtie2_large_index_ext=“.rev.1.bt2l”
This appears to have resolved the issue.