I’m using Humann3(humann v3.9) and metaphlan(MetaPhlAn version 4.1.0 (23 Aug 2023)) for feature enrichment.The data source I use is one of the fastq.gz files officially provided by humann3（SRS014476-Supragingival_plaque.fasta.gz）. Unfortunately, no error was reported during my operation. The output ‘SRS014476-Supragingival_plaque_pathabundance.tsv’ and ‘SRS014476-Supragingival_plaque_pathcoverage.tsv’ showed that there was no path. How to resolve this situation?I’ll attach the log file below. Thanks.
Pathway SRS014476-Supragingival_plaque_Abundance
UNMAPPED 0.0000000000
UNINTEGRATED 0.0000000000
Pathway SRS014476-Supragingival_plaque_Coverage
UNMAPPED 1.0000000000
UNINTEGRATED 1.0000000000
log file:
SRS014476-Supragingival_plaque.log.txt (89.7 KB)
If this is one of the down-sampled HMP demo files it is probably too shallow to confidently cover any pathways. The main demo input for HUMAnN has been engineered to cover some pathways despite being shallow.
thank you for your reply. What parameters can reduce the enrichment requirements and enrich some pathways in the presence of fewer microorganisms?
The issue I’m referring to has more to do with sequencing depth. If you are only covering 10% of a genome with reads (for example), then you’ll miss a lot of the genes that are needed to satisfy the reactions in metabolic pathways. You can compensate for this somewhat by gap-filling missing reactions or reducing HUMAnN’s database sequence coverage requirement, but at the end of the day it is not feasible to confidently reconstruct pathways from very low depth samples / very weakly covered species.