Identifying optimized values for m and s

[ktmbiome]

Hi!

I'm interested in using kaiju to identify fungi in my metagenomic short reads. However, when I use the default parameters, I get essentially the same profile of fungi across all samples, and when I expand the m and s parameters to try to reduce my false positive findings, I get no matches. Are there ways to know or optimize these parameters for different datasets?

[pmenzel]

Hi,
I think it's easier to adjust the E-value threshold using option -E instead of fiddling with the other two. This is more or less the same formula as used by NCBI BLAST.

Do you miss fungi that are actually contained in your samples? The RefSeq fungi collection is not super big, so the correct references might actually be missing in your kaiju database..