I recently downloaded this and updated all reference databases
I am running the following command after running kneaddata to remove human sequences
code:humann --input sample.r1_kneaddata.fastq --output ./humann3 --threads 24
error: flatpak: symbol lookup error: /lib64/libk5crypto.so.3: undefined symbol: EVP_KDF_ctrl, version OPENSSL_1_1_1b
/users/.conda/envs/humann3/lib/python3.12/site-packages/humann/quantify/MinPath12hmp.py:804: SyntaxWarning: invalid escape sequence ‘\d’
Followed by repeating errors: users/.conda/envs/humann3/lib/python3.12/site-packages/humann/quantify/MinPath12hmp.py:804: SyntaxWarning: invalid escape sequence ‘\d’
and finally: m = re.match(‘[1]+(?P\d+)’, aline)
Which repeats many many times prior to running metaphlan, bowtie2, diamond and then creating the genefamilies.tsv, pathabundance.tsv, and pathcoverage.tsv
Should I be worried? I’ve run humann several times in the past with older versions without this error.
Hi @gs5622, I figured out a way to install humann by using pip and using the ‘no-binary’ all flag. And then I installed the V22 chocophlan database after installing metaphlan separately using pypi. Most of the way that I installed it came from the instructions from humann4 but adapted for the older version of humann (not specifying the humann4 version). It worked for me without the errors. I was able to get humann to run to the end previously with the above error you also got, but it seems that I could not trust the files as things looked different when I got the end results with the non-error version of humann. If you need instructions for installing humann via pypi that I adapted here is the google doc: HUMAnN 4.0.0.alpha.1 Release Notes - Google Docs goodluck!
Thank you for sharing this. So far, I was able to get around the previous errors from Humann which were related to my Metaphlan database. Apparently, the bowtie2 index that I downloaded from the Metaphlan program directly is not the same as getting the 'full" database from humann directly. The warnings you shared still show up but the program seems to run to completion without error.
@gs5622 are you running kneaddata first to remove human sequences? This is where the bowtie2 database is utilized. Otherwise I updated all databases using the instructions at the bottom of humann3 – The Huttenhower Lab exactly as written on the instructions
It’s due to a change in Python’s regex engine in v3.12. If you install HUMAnN in an environment based on an earlier Python version it should resolve this error.