RAM in humann_regroup_table, etc. commands

Microbial community profiling HUMAnN
1

Hi! I have a very big dataset with thousands of samples and I am processing the Humann3 outputs with the commands that come with the 3.0.0 version. The renaming and regrouping take a huge amount of RAM and I was not able to perform the regrouping on non-stratified gene abundance tables with 800G of RAM. Is there any way to reduce the peak RAM requirements for the commands that process the joined humann3 output tables?

2

Hi! I am still wondering if there is a workaround to avoid having to use so much RAM for the humann_regroup table? Or can I just regroup the tables individually and use the humann_join_tables after?

3

Hello - Yes, that is a great idea. We too regroup the tables individually and then join them after. Thank you for the ping about the high memory usage for the regroup and join scripts. We have seen this too with very large files. We are looking into modifications to reduce the overall memory for these two scripts for very large files.

Thank you,
Lauren