I have the following conda environment
channels:
- conda-forge
- bioconda
dependencies:
- perl=5.22.0.1
- samtools=1.3
- kallisto=0.43.1
- cutadapt=1.9.1
- trim-galore=0.4.3
which gets loaded when I run snakemake --use-conda
Activating conda environment: /fullpathto/.snakemake/conda/4ac435d8
But then I get the error:
/usr/bin/bash: trim_galore: command not found
even though I can manually run the trim_galore executable successfully with: .snakemake/conda/4ac435d8/bin/trim_galore
The rule calling trim_galore is:
rule trim_galore:
input:
unpack(in_funcs.get_trim_galore_input(config))
output:
r1 = join(config['outs']['trim_fq_out_path'], '{sample}1_val_1.fq.gz'),
r2 = join(config['outs']['trim_fq_out_path'], '{sample}2_val_2.fq.gz'),
params:
out_path = config['outs']['trim_fq_out_path'],
conda:
join(config['protospork'], 'envs/biotools.yaml'),
shell:
'trim_galore --gzip -o {params.out_path} --paired {input.r1} {input.r2}'
Do I need to somehow specify that this trim_galore executable should be coming from the conda environment?
I figured it out, I was using a python installation from a virtualenv instead of making a conda environment. Hope this helps someone else
Prior workaround which is unsatisfying in many ways:
rule trim_galore:
input:
unpack(in_funcs.get_trim_galore_input(config))
output:
r1 = join(config['outs']['trim_fq_out_path'], '{sample}1_val_1.fq.gz'),
r2 = join(config['outs']['trim_fq_out_path'], '{sample}2_val_2.fq.gz'),
params:
exec_path = glob.glob(join(config['protospork'],'.snakemake','conda','*','bin','trim_galore'))[0],
out_path = config['outs']['trim_fq_out_path'],
conda:
join(config['protospork'], 'envs/biotools.yaml'),
shell:
'{params.exec_path} --gzip -o {params.out_path} --paired {input.r1} {input.r2}'