Snakemake fails to find conda in PBS

Question

[yixinzeng]

I tried to run snakemake on the PBS cluster, but snakemake reported an error as follows, the home directory is hidden with $HOME

$ snakemake --cluster "qsub -S /bin/sh -j y -cwd -pe thread 10-10 -q fat.q" -j 2 --core 10 --use-conda
Building DAG of jobs...
Traceback (most recent call last):
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/__init__.py", line 792, in snakemake
    success = workflow.execute(
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/workflow.py", line 1078, in execute
    dag.create_conda_envs(
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/dag.py", line 359, in create_conda_envs
    env.create(dryrun)
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/deployment/conda.py", line 393, in create
    pin_file = self.pin_file
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/common/__init__.py", line 217, in __get__
    value = self.method(instance)
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/deployment/conda.py", line 103, in pin_file
    f".{self.conda.platform}.pin.txt"
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/common/__init__.py", line 217, in __get__
    value = self.method(instance)
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/deployment/conda.py", line 96, in conda
    return Conda(
  File "$HOME/micromamba/envs/snakemake/lib/python3.8/site-packages/snakemake/deployment/conda.py", line 657, in __init__
    self.prefix_path = self.info["conda_prefix"]
KeyError: 'conda_prefix'
(snakemake)

The snakemake file is as follows

# config
configfile: "config.yaml"

# samples
SAMPLES = config["samples"]

# target rule
rule all:
    input:
        expand("mapped/{sample}.bam", sample=SAMPLES.keys())
# STAR mapping
rule star_map:
    input:
        reads = lambda wildcards: (SAMPLES[wildcards.sample]['r1'], SAMPLES[wildcards.sample]['r2'])
    output:
        "mapped/{sample}.bam"
    params:
        index = "$HOME/1-rna_editing/ref/STAR_genome_index",
        prefix = "mapped/{sample}"
    conda: "environment.yml"
    threads: 8
    shell:
        """
        STAR --runMode alignReads \
             --runThreadN {threads} \
             --genomeDir {params.index} \
             --readFilesCommand zcat \
             --outFileNamePrefix {params.prefix} \
             --outSAMtype BAM SortedByCoordinate \
             --readFilesIn {input.reads}
        """

This seems to be because snakemake did not find conda, but I did install micromamba.

$ which micromamba
$HOME/bin/micromamba

The version of snakemake is 7.32.4.

I found some similar reports: https://github.com/snakemake/snakemake/issues/1204

Any information would be very helpful, thank you so much.

Answer 1

[tim.booth]

Snakemake does not have support for micromamba. There is an open issue: https://github.com/snakemake/snakemake/issues/2322