7 EXTRA: Running nf-core/differentialabundance

We will use pipeline release v1.5, which takes the count matrix and sample metadata produced by nf-core/rnaseq and performs differential expression analysis using DESeq2.

To process the data, run the following command:

nextflow run nf-core/differentialabundance \
    -r 1.5.0 \
    -profile docker \
    -params-file params_degs.json \
    --outdir results/differentialabundance

The Nextflow command is stored in a bash script and can be executed by running:

bash 01_scripts/nfcore_differential_abundance.sh

The processing time is about 5 minutes.

7.0.1 Key parameters

The pipeline is configured via params_degs.json. The most important parameters we set are described below.

Inputs

• input — sample sheet linking each sample to its condition
• matrix — raw gene count matrix from Salmon (salmon.merged.gene_counts.tsv)
• transcript_length_matrix — gene length matrix for TPM normalisation
• contrasts — CSV file defining which groups to compare (treatment vs control)
• gtf — genome annotation file for E. coli K-12 MG1655

Feature annotation

• features_id_col: "gene_id" — locus tags (e.g. b0001) are used as the primary identifier
• features_name_col: "gene_name" — gene symbols (e.g. thrL) are used as display names
• features_gtf_feature_type: "transcript" — annotations are extracted at transcript level

Filtering — genes must pass all three criteria to be retained:

• filtering_min_abundance: 10 — minimum count of 10
• filtering_min_samples: 2 — present in at least 2 samples
• filtering_grouping_var: "group" — filtering applied per experimental group

DESeq2

• deseq2_fit_type: "local" — uses a local dispersion fit, more appropriate for smaller sample sizes than the default parametric fit

Differential expression threshold

• differential_min_fold_change: 1.0 — only genes with |FC| ≥ 1.5 are reported as DE

Disabled modules

• gprofiler2_run: false and gsea_run: false — the pipeline’s built-in functional enrichment steps are turned off because to save time here.