Simple snakemake pipelines for common/standard bioinformatics analyses

This repo contains a collection of ad-hoc and flexible pipelines for running common bioinformatics tools on a directory of input files.

See Available Pipelines section for a list & description of available workflows. Well defined/flexible pipelines include:

• Salmon transcript quantification
• Pull FASTQs from coordinate-sorted BAM files
• Generate summary statistics from BAM files with samtools stats
• Infer strandedness of RNA-seq reads from BAM files with infer_experiment.py

Running instructions

Perform a dry-run to check run declaration:

snakemake -n -p -s single_steps/<workflow>.smk

Available pipelines

Salmon

Run Salmon transcript quantification on a directory of input FASTQ files (paired-end or single-end). The workflow can use a pre-provided index or generate an full index with the genome sequence as decoys from provided reference files.

If generating a custom index, you have the following options:

• Provide a GTF file of transcript models - transcript sequences will be extracted using gffread
• Provide a FASTA file of transcript sequences - will be passed straight to salmon index

If generating a custom index, you must provide a genome sequence FASTA file. To use a pre-computed index, you just need to provide the workflow with the path to the directory containing the index.

Snakefile: single_steps/salmon.smk

Config file: config/salmon_config.yaml

Cluster config file: config/cluster/salmon.yaml

Pull FASTQs from BAM files

Uses samtools to sort input coordinate-sorted BAM files by read name and extract reads to FASTQ files. Supports single-end or paired-end reads.

Note that name sorting is performed using samtools collate (standard mode). This ensures that reads of the same name are grouped together in the BAM file, but does not perform a full alphabetical sort (and so is quicker than samtools sort). This means that read pairs will be ordered randomly in the output BAM file. See documentation for a full breakdown and description of potential ramifications (for our typical RNA-seq analyses this shouldn't be an issue).

Snakefile: single_steps/sort_pull.smk

Config file: config/sort_pull_config.yaml

Cluster config file: config/cluster/sort_pull.yaml

Generate summary statistics from BAM files with samtools stats

Runs samtools stats over a set of input BAM files. Also collapses the 'summary/SN section' into a single summary table for all samples. See documentation for a full breakdown/description of calculated metrics.

Snakefile: single_steps/samtools_stats.smk

Config file: config/samtools_stats_config.yaml

Cluster config file: config/cluster/samtools_stats.yaml

Note: Requires pandas to be installed outside of pipeline, which is usually satisfied by a standard snakemake installation.

Infer strandedness of an RNAseq experiment from BAM files with infer_experiment.py

Runs RSeQC's infer_experiment.py over a set of input BAM files to infer the 'strandedness' of the input RNA reads. Also requires transcript annotation in BED12 format, which can be generated from a GTF file using the recipe in single_steps/gtf_to_bed12.smk.

See documentation for interpretation of the output files. The following blog posts are also handy for translating the definition to the correct parameter for popular RNA-seq tools - 1, 2.

Snakefile: single_steps/infer_experiment.smk

Config file: config/infer_experiment_config.yaml

Cluster config file: config/cluster/infer_experiment.yaml

Command to submit to UCL cluster: source submit.sh infer_experiment <optional_run_name>