I have been trying fo 1 month to get humann 4 alpha to work. eventually i have managed to get it to work using the metaphlan v4.0.6 as long as i run metaphlan separately and then use humann referencing to the metaphlan.tsv.
The next work around seems to be that now DIAMOND is reverting to using 1 thread.
To fix that I am guessing I have to tell humann to let DIAMOND use more threads.
export HUMANN_DIAMOND_OPTIONS=“–threads 16” for example. And i am guessing that its the same for bowtie3 and metaphlan.
Humann3 let you do: humann --input file --output directory --threads 16 and go and have lunch. Now I am finding with version 4 alpha i have to set up so much more. I have to tell metaphlan where its database is, its index, tell it to add virures and taxonomy all and finally stop it trying to download the chocophlan database every time.
metaphlan test.fastq
–input_type fastq
–bowtie2db /path/to/metaphlan_db/
–index mpa_vOct22_CHOCOPhlAnSGB_202403
–output test_metaphlan.tsv
–add_viruses
–tax_lev a
–offline #to stop it trying to download the database everytime)
humann
–input test.fastq
–output humann_out
–taxonomic-profile test_metaphlan.tsv #because humann 4 no longer can run metaphlan and carry on to humann with the file. it throws a database error.
Could you please update the installation and faqs for this as this would be helpful for all those who come after. This is a great pipeline but its getting less and less usable and i want to be able to use custom databases in future and its not looking good. I am not a coder or bioinformaticist I am a biologist who can read and type a bit…so any help you can give people like me is not wasted!