Execution of pipelines

Next to the ZARP workflow for RNA-Seq analysis, this project comes with two auxiliary workflows for fetching samples from the Sequence Read Archive and populating a sparse sample table with inferred sample metadata. This section describes how to run each of these workflows.

Prerequisites

All usage examples in this section assume that you have already installed ZARP.

How to analyze your RNA-Seq samples?

1. Assuming that your current directory is the workflow repository's root directory, create a directory for your workflow run and move into it with:

   mkdir config/my_run
   cd config/my_run
   

2. Create an empty sample table and a workflow configuration file:

   touch samples.tsv
   touch config.yaml
   

3. Use your editor of choice to populate these files with appropriate values. Have a look at the examples in the tests/ directory to see what the files should look like, specifically:

   • samples.tsv
   • config.yaml

4. Create a runner script. Pick one of the following choices for either local or cluster execution. Before execution of the respective command, you need to remember to update the argument of the --singularity-args option of a respective profile (file: profiles/{profile}/config.yaml) so that it contains a comma-separated list of all directories containing input data files (samples and any annotation files etc) required for your run.

   Runner script for local execution:

   cat << "EOF" > run.sh
   #!/bin/bash
   
   snakemake \
       --profile="../../profiles/local-singularity" \
       --configfile="config.yaml"
   
   EOF
   

   OR

   Runner script for Slurm cluster execution (note that you may need to modify the arguments to --jobs and --cores in the file: profiles/slurm-singularity/config.yaml depending on your HPC and workload manager configuration):

   cat << "EOF" > run.sh
   #!/bin/bash
   mkdir -p logs/cluster_log
   snakemake \
       --profile="../profiles/slurm-singularity" \
       --configfile="config.yaml"
   EOF
   

   Note: When running the pipeline with Conda you should use local-conda and slurm-conda profiles instead.

   Note: The slurm profiles are adapted to a cluster that uses the quality-of-service (QOS) keyword. If QOS is not supported by your slurm instance, you have to remove all the lines with "qos" in profiles/slurm-config.json.

5. Start your workflow run:

   bash run.sh
   

6. To find out more information on the output files please go to the output files section.

How to fetch sequencing samples from SRA?

An independent Snakemake workflow workflow/rules/sra_download.smk is included for the download of sequencing libraries from the Sequence Read Archive and conversion into FASTQ.

The workflow expects the following parameters in the configuration file:

• samples, a sample table (tsv) with column sample containing SRR identifiers (ERR and DRR are also supported), as in this example samples.tsv file.
• outdir, an output directory
• samples_out, a pointer to a modified sample table with the locations of the corresponding FASTQ files
• cluster_log_dir, the cluster log directory.

For executing the example with Conda environments, one can use the following command (from within the activated zarp Conda environment):

snakemake --snakefile="workflow/rules/sra_download.smk" \
          --profile="profiles/local-conda" \
          --config samples="tests/input_files/sra_samples.tsv" \
                   outdir="results/sra_downloads" \
                   samples_out="results/sra_downloads/sra_samples.out.tsv" \
                   log_dir="logs" \
                   cluster_log_dir="logs/cluster_log"

Alternatively, change the argument to --profile from local-conda to local-singularity to execute the workflow steps within Singularity containers.

After successful execution, results/sra_downloads/sra_samples.out.tsv should contain:

sample  fq1     fq2
SRR18552868     results/sra_downloads/compress/SRR18552868/SRR18552868.fastq.gz 
SRR18549672     results/sra_downloads/compress/SRR18549672/SRR18549672_1.fastq.gz       results/sra_downloads/compress/SRR18549672/SRR18549672_2.fastq.gz
ERR2248142      results/sra_downloads/compress/ERR2248142/ERR2248142.fastq.gz 

How to infer sample metadata?

An independent Snakemake workflow workflow/rules/htsinfer.smk that populates the samples.tsv required by ZARP with the sample specific parameters seqmode, f1_3p, f2_3p, organism, libtype and index_size. Those parameters are inferred from the provided fastq.gz files by HTSinfer.

Note: The workflow uses the implicit temporary directory from snakemake, which is called with [resources.tmpdir].

The workflow expects the following config:

• samples, a sample table (tsv) with column sample containing sample identifiers, as well as columns fq1 and fq2 containing the paths to the input fastq files see example here. If the table contains further ZARP compatible columns (see pipeline documentation), the values specified there by the user are given priority over htsinfer's results.
• outdir, an output directory
• samples_out, path to a modified sample table with inferred parameters
• records, set to 100000 per default

For executing the example one can use the following (with activated zarp environment):

cd tests/test_htsinfer_workflow
snakemake \
    --snakefile="../../workflow/rules/htsinfer.smk" \
    --restart-times=0 \
    --profile="../../profiles/local-conda" \
    --config outdir="results" \
             samples="../input_files/htsinfer_samples.tsv" \
             samples_out="samples_htsinfer.tsv" \
             log_dir="logs" \
             cluster_log_dir="logs/cluster_log" \
    --notemp \
    --keep-incomplete

However, this call will exit with an error, as not all parameters can be inferred from the example files. The argument --keep-incomplete makes sure the samples_htsinfer.tsv file can nevertheless be inspected.

After successful execution - if all parameters could be either inferred or were specified by the user - [OUTDIR]/[SAMPLES_OUT] should contain a populated table with parameters seqmode, f1_3p, f2_3p, organism, libtype and index_size.