UniOP: Universal operon prediction for high-throughput prokaryotic (meta-)genomics data

UniOP was implemented in Python 3.11.5 and runs on MacOS or Linux systems.

Overview

Operon prediction remains challenging for poorly characterized prokaryotic genomes, particularly for metagenome-assembled genomes (MAGs) that lack functional annotations and evolutionary relatives. Existing methods typically rely on supervised training, comparative genomics, or external databases, limiting their applicability to such data.

UniOP address this gap by inferring operon structure directly from intergenic distance distributions within the target genome. It requires no external information, no training data, and no species-specific assumptions.

UniOP takes as input either a nucleotide genome sequence (FASTA), a protein sequence file (FAA), or a gene annotation file (GFF), and outputs pairwise operon probabilities and assembled operon units. The tool is implemented in Python and runs on macOS or Linux systems.

How to use UniOP

Dependencies

UniOP requires the following:

* Python (version 3.11.5 recommended; the results in the paper were obtained with this version)
* Python libraries: argparse, pandas, numpy, scikit-learn, datetime
* Prodigal (required only when using nucleotide FASTA as input)

You can install Prodigal via:

conda install -c bioconda prodigal

Quickstart

Clone the repository and run UniOP on the provided demo data.

git clone https://github.com/hongsua/UniOP.git
cd UniOP/src

Using a genomic sequence (auto-runs Prodigal)

python UniOP -i ../demo/GCF_000005845.2.fna --bin_dir /path/to/prodigal

Using a FAA file

python UniOP -a ../demo/GCF_000005845.2.faa --faa_source prodigal_faa   # for Prodigal FAA
python UniOP -a ../demo/GCF_000005845.2.faa --faa_source ncbi_faa   # for NCBI FAA

Using a GFF file

python UniOP -f ../demo/GCF_000005845.2.gff --gff_source prodigal_gff   # for Prodigal GFF
python UniOP -f ../demo/GCF_000005845.2.gff --gff_source ncbi_fgff   # for NCBI GFF

Note: If you use the -i option, Prodigal must be installed and accessible via --bin_dir.

This will output:

• uniop.pred -pairwise operon probabilities for all adjacent same-strand gene pairs.
• uniop.operon -assembled operon units (if not disabled with --no_operon_assembly)

By default, output files are written to the same directory as the input file. You can specify an output folder with -t:

python UniOP -i ../demo/GCF_000005845.2.fna -t ./results --bin_dir /path/to/prodigal

Supported input formats

Input type	Option	Required argument	Example command
Genomic sequence (FNA)	-i	--bin_dir (if not in PATH)	python UniOP.py -i genome.fna --bin_dir /usr/bin
Prodigal FAA	-a	--faa_source prodigal_faa	python UniOP.py -a genes.faa --faa_source prodigal_faa
NCBI FAA	-a	--faa_source ncbi_faa	python UniOP.py -a genes.faa --faa_source ncbi_faa
Prodigal GFF	-f	--gff_source prodigal_gff	python UniOP.py -f genes.gff --gff_source prodigal_gff
NCBI GFF	-f	--gff_source ncbi_gff	python UniOP.py -f genes.gff --gff_source ncbi_gff

Full parameter list

python UniOP --help

Citation

If you use UniOP in your research, please cite the preprint:

Su, H., Zhang, R., & Söding, J. (2024). UniOP: a universal operon prediction for high-throughput prokaryotic (meta-) genomic data using intergenic distance. bioRxiv, 2024-11.

We will update the official citation once the manuscript is published.

Support

If you have questions, encounter any bugs, or need help applying UniOP to your data, please open an issue on GitHub or contact us directly:

• GitHub Issues: https://github.com/hongsua/UniOP/issues
• hong.su[at]nankai.edu.cn

We welcome feedback, feature requests, and community contributions.