Error when using snakemake with RSEM

Question

[nattzy94]

I am trying to run RSEM on a number of samples using snakemake. However, when I run it on the cluster with snakemake --use-conda --cluster "qsub -pe smp 12" -j 10 -s ./rsem_snakefile, I get this error:

Activating conda environment: /Nathaniel/snakemake-tutorial/.snakemake/conda/dcb8dd66
/bin/bash: /Nathaniel/snakemake-tutorial/.snakemake/conda/dcb8dd66/bin/rsem-calculate-expression -p 12 --paired-end --star-gzipped-read-file /Nathaniel/raw_data/rna_seq/m19-32h_1.fastq.gz /Nathaniel/raw_data/rna_seq/m19-32h_2.fastq.gz --star --star-path /home/STAR-2.7.3a/bin/Linux_x86_64_static /Nathaniel/customgtf_rnaseq/ensemblgtf_results/rsem/ensembl99 --output-genome-bam /Nathaniel/rsem_snakemake/m19-32h/m19-32h > /Nathaniel/rsem_snakemake/m19-32h/m19-32h_rsem.log: No such file or directory

My rsem_snakefile is as follows:

import json
from os.path import join, basename, dirname

configfile: '/Nathaniel/snakemake-tutorial/config.yml'

OUT_DIR = config['OUT_DIR']
FILES = json.load(open(config['SAMPLES_JSON']))
SAMPLES = sorted(FILES.keys())

rule all:
        input:
                [OUT_DIR + "/" + x for x in expand('{sample}/{sample}.transcript.bam', sample = SAMPLES)]

rule rsem:
        input:
                r1 = lambda wildcards: FILES[wildcards.sample]['R1'],
                r2 = lambda wildcards: FILES[wildcards.sample]['R2']
        output:
                join(OUT_DIR,'{sample}','{sample}.transcript.bam')
        conda:
                '/Nathaniel/rsem_snakemake/rsem.yaml'
        log:
                '/Nathaniel/rsem_snakemake/{sample}/{sample}_rsem.log'
        shell:
                """
                "/Nathaniel/snakemake-tutorial/.snakemake/conda/dcb8dd66/bin/rsem-calculate-expression -p 12 --paired-end --star-gzipped-read-file {input.r1} {input.r2} --star --star-path /STAR-2.7.3a/bin/Linux_x86_64_static /Nathaniel/customgtf_rnaseq/ensemblgtf_results/rsem/ensembl99 --output-genome-bam /Nathaniel/rsem_snakemake/{wildcards.sample}/{wildcards.sample} > {log}"
                """

config.yaml looks like this:

OUT_DIR: '/Nathaniel/rsem_snakemake'
SAMPLES_JSON: '/Nathaniel/snakemake-tutorial/samples.json'

First few lines of samples.json:

{
    "16hr-33": {
        "R1": [
            "/Nathaniel/raw_data/rna_seq/16hr-33_1.fastq.gz"
        ],
        "R2": [
            "/Nathaniel/raw_data/rna_seq/16hr-33_2.fastq.gz"
        ]
    },
    "32hr-25": {
        "R1": [
            "/Nathaniel/raw_data/rna_seq/32hr-25_1.fastq.gz"
        ],
        "R2": [
            "/Nathaniel/raw_data/rna_seq/32hr-25_2.fastq.gz"
        ]
    },

Could this be a problem where the cluster is unable to activate the Conda environment? I can't figure this out.

Answer 1

[nattzy94]

Figured out what was wrong, the shell command in rsem snakefile should not have quotes around them. It should read:

shell:
                """
                /Nathaniel/snakemake-tutorial/.snakemake/conda/dcb8dd66/bin/rsem-calculate-expression -p 12 --paired-end --star-gzipped-read-file {input.r1} {input.r2} --star --star-path /STAR-2.7.3a/bin/Linux_x86_64_static /Nathaniel/customgtf_rnaseq/ensemblgtf_results/rsem/ensembl99 --output-genome-bam /Nathaniel/rsem_snakemake/{wildcards.sample}/{wildcards.sample} > {log}
                """