Hi All,
I would appreciate clarification on whether KneadData is recommended for mouse fecal samples that underwent shotgun metagenomic sequencing. My understanding is that fecal samples generally contain less host DNA than tissue-derived samples, but I’m not sure whether host read removal is still necessary before running MetaPhlAn and HUMAnN. Because KneadData can be computationally expensive at scale, if this step is not essential for mouse fecal metagenomes, we are considering omitting it.
Thanks!
Claire
Generally we recommend running kneaddata for fecal samples since there is still some host DNA. However, this is less important for mouse samples because one of the main reasons to remove host DNA is to de-identify samples for the privacy of human participants. For metaphlan, there probably is little difference in output between running kneaddata and not. For humann, the translated search step might give you some issues and even if the output ends up the same, it may take much longer to run (so much so that you gain time by running kneaddata in some cases).
Ultimately it is up to you, and you may want to run a couple samples both ways to compare run-times and results. We’d also recommend running trimmomatic regardless of if you run the full kneaddata suite.
1 Like
Hi! We proceeded with KneadData for our analysis. Thank you for your input - we appreciate it.
1 Like