There was a bug caused by the .bashrc source directive at the
start of the snakefile
On a clean Ubuntu 18.04 container envs were being activated for each
rule but then seemingly deactivated when snakemake started to execute
the shell component.
This caused the python2.7 reliant phylophlan and checkm test scripts to
error.
For example, output from adding conda env list to the test_checkm
rule shows the correct activation before the shell but the wrong activated
environment (shown by asterisk) in the shell instance for the rule:
activating conda environment: /home/root/MAGpy/.snakemake/conda/377a289b
/home/root/MAGpy/.snakemake/conda/377a289b
/home/root/MAGpy/.snakemake/conda/8a832417
base
* /root/miniconda3
I think the directive means snakemake (v5.11.2) was sourcing the
.bashrc when starting the shell for each rule and resetting to the
base environment before executing the rule script commands.
Without that directive the code everything seems to run fine for the
tests and full pipeline when tested with snakemake version >5
Otherwise, it was a few trivial updates to documentation, and adding a command to set the checkm dataroot as it was waiting for input and hanging on the test_checkm rule.
There was a bug caused by the
.bashrc
source directive at the start of the snakefileOn a clean Ubuntu 18.04 container envs were being activated for each rule but then seemingly deactivated when snakemake started to execute the shell component.
This caused the python2.7 reliant phylophlan and checkm test scripts to error.
For example, output from adding
conda env list
to thetest_checkm
rule shows the correct activation before the shell but the wrong activated environment (shown by asterisk) in the shell instance for the rule:I think the directive means snakemake (v5.11.2) was sourcing the
.bashrc
when starting the shell for each rule and resetting to the base environment before executing the rule script commands.Without that directive the code everything seems to run fine for the tests and full pipeline when tested with snakemake version >5
Otherwise, it was a few trivial updates to documentation, and adding a command to set the checkm dataroot as it was waiting for input and hanging on the
test_checkm
rule.