Unnecessary compleasm failure alert

[juliadouglasf]

Hello,

I have been running BRAKER using a conda environment coupled with GeneMark and ProtHint git installations. I know this is not the recommended method, but it is simpler, especially if I incorporate it into a Snakemake pipeline later on in my work.

I received a $COMPLEASM_PATH not set! error even though both compleasm and best_by_compleasm.py are included in my conda environment bin path. I think BRAKER is checking for compleasm.py, which doesn't seem to exist. With my conda environment activated, compleasm --help works as expected. Since I believe BRAKER only uses the best_by_compleasm.py script, I don't think this $COMPLEASM_PATH not set! error should be thrown and stop my execution of BRAKER.

[KatharinaHoff]

best_by_compleasm.py calls compleasm.py . It has a command line argument --compleasm_bin that needs to hand over the compleasm.py file. If your compleasm is hidden in an environment that is not activitated while BRAKER is running, then BRAKER cannot see compleasm.py, and can thus not determine the $COMPLEASM_PATH variable that points to the location of compleasm.py .

[juliadouglasf]

I have compleasm in the same environment as BRAKER, so I don't think I will have the issue of compleasm not being found while BRAKER is running. The compleasm command seems to be working the same as compleasm.py.

[aforestsomewhere]

I believe I was able to get around this by creating a symlink to the compleasm within my conda installation of BRAKER3, as it seemed to strictly require compleasm.py (within my $CONDA_PREFIX/braker-3.0.8-0/bin: ln -s compleasm compleasm.py)

Edit: actually this caused a circular import error later in the pipeline. But downgrading compleasm to version 0.2.5 and modifying all calls to "compleasm.py" to "compleasm" in the main braker.pl script allowed the pipeline to run to completion.