No MetaPhlAn BowTie2 database found (--index option)! Humann3.6

Hi! I know there are other open questions on this topic, but as I am using the latest version of Humann (3.6) I wanted to open a new issue. I am currently going through the setup process of Humann in a conda environment, and followed the steps on your installation guide and ran the following commands:

conda create --name biobakery3 python=3.7
conda config --add channels defaults
conda config --add channels bioconda
conda config --add channels conda-forge
conda config --add channels biobakery
conda install humann -c biobakery```

I then downloaded all the databases:

humann_databases --download chocophlan full /scratch4/jsuez1/week12/chocophlan --update-config yes
humann_databases --download uniref uniref90_diamond/scratch4/jsuez1/week12/diamond --update-config yes
humann_databases --download utility_mapping full /scratch4/jsuez1/week12/diamond --update-config yes

Then when I try to run either a sample of mine or the demo.fastq.gz, i get this error:

Running metaphlan ........

CRITICAL ERROR: Error executing: /home/hhallow1/.conda/envs/biobakery3/bin/metaphlan /scratch4/jsuez1/week12_virome/sample_results/demo_humann_temp/tmpm6r5uqir/tmpe2u0gnbk -t rel_ab -o /scratch4/jsuez1/week12_virome/sample_results/demo_humann_temp/demo_metaphlan_bugs_list.tsv --input_type fastq --bowtie2out /scratch4/jsuez1/week12_virome/sample_results/demo_humann_temp/demo_metaphlan_bowtie2.txt

Error message returned from metaphlan :

No MetaPhlAn BowTie2 database found (--index option)!

Expecting location bowtie2db

Exiting...

I have tried deleting and re-installing the databases, i’ve tried updating the config file (it was already updated), i’ve tried completely re-installing humann in a new environment and im not sure what else to do. i’ve used humann in the past with no problems at all. any help would be appreciated! Thanks!

Hi! just wanted to update … I still am stuck on this step. It seems that when i download the database using

humann_databases --download chocophlan full /scratch4/jsuez1/week12/chocophlan --update-config yes

The file becomes unzipped. I tried manually combining all files into one .fna by doing this

cat chocophlan/* > mpa_v30_CHOCOPhlAn_201901.fna

then using bowtie-build to build my index

bowtie2-build mpa_v30_CHOCOPhlAn_201901.fna mpa_v30_CHOCOPhlAn_201901

I got about 1000 warnings from bowtie that each reference was empty, i.e contains all Ns rather than ATCG, before i killed the job

So my question is, why is the database unzipping when i download it? what is the correct why to cat these files together to be indexed?

I am running Metaphlan 4.0.2 and Humann 3.6

Thanks!

Hello!

Looks like you have issues with metaphlan databases.

Did you try to install MetaPhlan database separately?

metaphlan --install --bowtie2db your_path_to_db/MetaPhlan4/

Then run humann with index option:
--metaphlan-options "--index $index --bowtie2db your_path_to_db/MetaPhlan4"

where index is an index from MetaPhlan database installed (check the name of the file with “.tar” extention and copy it without extension).

Hi!

On my third try of going through the installation tutorial, it worked. not sure what was going on with it, but it seems like the metaphlan database was not installing the entire way. This last time though everything installed correctly and i can run humann.

thanks!

Glad this worked out. We’re in the process of revising the MetaPhlAn database download/index process to make it easier (and less resource-intensive) for users. With the switch to SGB markers in MetaPhlAn 4.0 the process is a lot more computationally expensive than it was in previous versions.