Swtiching from uniref90 to uniref 50 with only unclassified reads

Hi there!
I’ve been enjoying HUMAnN a lot so far, I just have one question.

My samples are from the guts of these tropical carrion feeding bees. I realized after running HUMAnN with uniref90 and default parameters that my samples probably are very uncharacterized and that I should have used uniref50.

Is there anyway to get a fastq of only the resulting “unclassified” reads and then run uniref50 on those? Or what do you think is the best way I should go about getting those classified? I have 95-99% unaligned reads after nucleotide and translated alignments. So, I’m just assuming it’s because I didn’t use uniref50 but is there any other reason you image that could be?


Hello Jess, Glad to hear you have been enjoying HUMAnN so far! There is a file generated of the unaligned reads after both the nucleotide and translated steps. However, if most are unaligned it might make sense to start with the original fastq file for the UniRef50 run.

Have you run the data set initially through a QC step like Kneaddata? If not, I would recommend that as a first step to see if you can improve the alignment rates.