Carveme Error calling stat on file /generated/bigg_proteins.dmnd

[jorgeln0]

Dear author,

I installed locally from the latest Github version and am trying to run carveme through the metaGEM pipeline. When calling carve, I have to specify the full python path (/usr4/spclpgm/jlopezna/.local/bin/carve). I was looking at previous issues that suggested running carve genome.faa -v and whenever I did this I received:

"Error: Error calling stat on file /usr4/spclpgm/jlopezna/.local/lib/python3.6/site-packages/carveme/data/generated/bigg_proteins.dmnd Failed to run diamond.

It seems like there is a missing file that carve is expecting. It might be worth noting that I am I working from a shared computing cluster network but that hasn't caused any issues in the past. Any suggestions?

[cdanielmachado]

It seems that something went wrong with the installation, since you are missing one of the internal files. Also, you should be able to execute carveme without specifying the full path.

Did you install using pip ? Can you try doing a clean install ?

pip uninstall carveme
pip install carveme

Make sure there are no errors or warnings during the installation.

[franciscozorrilla]

Hello, happened to see this issue by chance, just wanted to point out that metaGEM does indeed use pip install carveme --user to install CarveMe, although I have not encountered the error reported by @jorgeln0 . I found that pip installations fail on some clusters without the --user flag.

https://github.com/franciscozorrilla/metaGEM/wiki/Manual-setup

[cmfield]

If you look at setup.py the paths to some of the files in data are just wrong, so they aren't installed. There's no bigg_proteins.faa in my data/generated directory for instance, so no diamond database can be made.

[cmfield]

I've attached a patch to remedy this. CarveMe-1.5.2.patch