Comparative Genomics Pipeline

This repository contains the analysis code accompanying the manuscript:

Dahal, S., Liew, C. S., Riethoven, J.-J., & Rojas, C. M. (2026). Comparative genomics approaches to identify genomic regions associated with the antimicrobial activity of Pseudomonas protegens PBL3. Microbial Genomics, 12(9). https://doi.org/10.1099/mgen.0.001827

This repository covers the hard-to-reproduce post-processing steps for two independent comparative genomics methods — anvi'o pangenomics and progressiveMauve whole-genome alignment — used to identify genes and genomic regions present in all positive-group genomes but absent from all negative-group genomes. Results from both methods are cross-referenced and merged into a unified report.

Note: This code is provided for reproducibility purposes only. It is not actively maintained and issues or pull requests are not monitored.

---

Requirements

• Python ≥ 3.10
• pandas ≥ 2.0
• numpy ≥ 1.24
• openpyxl ≥ 3.1

Install dependencies:

pip install -r requirements.txt

---

Input files

Place input files in the data/ directory (real data) or use example_data/ to test with the provided anonymised dataset.

File	Description
genomes_metadata.tsv	Genome names, activity labels (positive/negative), and backbone column positions
gc_occurrence.tsv*	Gene cluster × genome count matrix from anvi'o pangenome analysis
pangenome_gene_clusters_summary.tsv	Per-gene anvi'o gene cluster membership and amino acid sequences
mauve_backbone.backbone	progressiveMauve backbone output (locally collinear blocks)
reference_genome_annotations.tsv	Gene coordinates and descriptions for the reference genome (GTF-derived)
reference_prodigal_annotations.tsv	Prodigal gene calls for the reference genome with COG/KO annotations

*gc_occurrence.tsv is generated following the instructions here.

genomes_metadata.tsv format

genome          activity    backbone_position
strain_pos_ref  positive    0
strain_pos_2    positive    1
...
strain_neg_1    negative    4

• backbone_position maps each genome to its column pair in the Mauve backbone file (position 0 = reference genome, columns 0–1; position N = columns 2N–(2N+1)).
• The genome with backbone_position == 0 is used as the reference throughout the pipeline.

---

How to run

Run scripts in order. All scripts accept --data and --results arguments to specify input and output directories (defaults: data/ and results/).

# Using real data
python 01_process_anvio.py
python 02_process_mauve.py
python 03_integrate_results.py
python 04_merge_results.py

# Using example data
python 01_process_anvio.py --data example_data
python 02_process_mauve.py --data example_data
python 03_integrate_results.py --data example_data
python 04_merge_results.py

---

Output files

All output files are written to results/.

Script	Output file	Description
01	gc_exclusive_to_positive_annotated.tsv	Gene-level details for gene clusters exclusive to the positive group (reference genome)
02	mauve_gene_level.tsv	Genes in LCBs exclusive to the positive group, with reference genome annotations
03	mauve_gene_level_with_anvio.tsv	Mauve genes annotated with COG/KO and anvi'o GC membership flag
03	gc_exclusive_with_mauve_flag.tsv	anvi'o exclusive GC genes with Mauve overlap flag, coordinates, and COG/KO annotations
04	merged_results.xlsx	Unified report combining both methods; each gene labelled both, mauve, or anvio

Expected row counts (example_data run)

Output file	Expected rows
gc_exclusive_to_positive_annotated.tsv	98
mauve_gene_level.tsv	223
mauve_gene_level_with_anvio.tsv	223
gc_exclusive_with_mauve_flag.tsv	98
merged_results.xlsx	188 (87 mauve, 74 both, 27 anvio)

---

Example data

The example_data/ directory contains anonymised versions of the real input files for testing. Genome names have been replaced with generic labels (strain_pos_ref, strain_pos_2, …, strain_neg_1, …), and gene identifiers with sequential labels (gene_00001, …). Large files are subsampled to include all biologically relevant rows (the exclusive gene clusters) plus 500 randomly selected additional rows.

To regenerate example_data/ from data/:

python make_dummy_data.py

---

AI assistance

Code in this repository was developed with assistance from Claude Code (Anthropic, claude-sonnet-4-6), used specifically for code refactoring, documentation, and repository organization.