PSOSP: Prophage SOS-dependency Predictor

PSOSP (Prophage SOS dependency Predictor) is a novel bioinformatics tool to predict prophage induction modes by analyzing the heterology index (HI) of LexA protein binding to target DNA, classifying prophages into SOS-dependent (SdPs) and SOS-independent (SiPs).

Table of contents (Chinese Tutorial/中文说明✨)

• PSOSP: Prophage SOS-dependency Predictor
• Webserver
• Background
  • Principle
  • Workflow
  • Experimental validation
  • Significance for prophage isolation
• Instructions
  • Input requirements
  • Dependencies
  • Installation
  • Input files
  • How to run
  • Outputs
• Citation

Webserver

We provide an online platform (PSOSP) for rapid prediction of bacteriophage induction modes: https://vee-lab.sjtu.edu.cn/PSOSP/. There you can upload your host and phage genomes and get the prediction results.

Background

Principle

Temperate phages integrate into the bacterial host genome as prophages. Under normal conditions, the LexA protein binds to the SOS box within the prophage, repressing the expression of phage-related genes and maintaining the lysogenic state. Upon external stimuli (such as exposure to DNA-damaging agents), the RecA protein is activated, leading the self-cleavage of LexA and its dissociation from the SOS box. This relieves the prophage repression, triggering the temperate phage to enter the lytic cycle and thereby facilitating its proliferation.

Workflow

• LexA & Canonical SOS Box (CBS) Identification : Scanning the host genome to identify LexA protein and canonical SOS boxes (CSBs) located upstream of the lexA gene
• Heterology Index (HI) Calculation: Identifying potential SOS boxes (PSBs) across bacterial genomes, calculating the Heterology Index (HI) for each PSB and establishing classification thresholds (HI_c1 and HI_c2) via Mean Shift clustering results
• PSB scan in prophage: Scanning PSBs within prophage promoter regions and determining of the minimum HI (HI_min)
• Prophage categoriation : Evaluating the ability of LexA binding to prophage promoter regions by comparing HImin with thresholds
  • HI_min ≤ HI_c1 → SdP (SOS-dependent Prophage)
  • HI_min ≥ HI_c2 → SiP (SOS-independent Prophage)
  • HI_c1 < HI_min < HI_c2 → SuP (SOS-uncertain Prophage)

Experimental validation

We have validated PSOSP’s accuracy using 14 experimentally confirmed bacteriophages spanning 10 viral families (including 2 Peduoviridae, 3 Inoviridae, and 9 distinct novel families), with their hosts covering 7 bacterial genera (Salmonella, Escherichia, Vibrio, Pseudomonas, Serratia_J, Hafnia, and Shewanella) across 3 bacterial orders (Enterobacterales, Enterobacterales_A, and Pseudomonadales). Significantly, all PSOSP predictions for these bacteriophages showed complete consistency with experimental evidence, demonstrating the tool’s versatility and reliability across broad taxonomic ranges.

Significance for prophage isolation

We propose that future phage isolation efforts could first use PSOSP to determine phage type

• For SdPs, conventional SOS-inducing agents (e.g., MMC, UV) remain appropriate.
• For SiPs, SOS-independent inducers such as DPO, C4-HSL, EDTA, and pyocyanin, or physical factors like varying salinity, temperature, and pH, should be considered

Instructions

Input requirements

For host:

• Host Taxon Suitability: PSOSP is primarily suitable for Gammaproteobacteria, including multiple critical pathogens such as Vibrio cholerae, Pseudomonas aeruginosa, Yersinia pestis, Escherichia coli, Salmonella enterica, Shigella spp., and Klebsiella spp. Compatible bacterial genera can be viewed on the Statistics page of the PSOSP online website.
• Genome Quality: We advise using host genomes with a completeness score above 90%, since low-quality genomes may lose the LexA protein and lead to poor results. You can assess the completeness of your genome assembly using CheckM2.
• Multi-Contig Genomes: If the host genome consists of multiple contigs, ensure the input host genome file contains all contigs (i.e., provide the genome assembly as a single multi-contig file).

For prophage:

• Genome Quality: PSOSP utilizes CheckV for quality assessment. Predictions are reliable for prophages with >=90% CheckV completeness; accuracy decreases for completeness <90%. If you are certain your phage genome is complete, you may disregard the CheckV results in the output file (as CheckV assessments can occasionally be inaccurate).
• Multiple Inputs: The input phage genome file can contain sequences for multiple prophages.
• Host Association: Input phages must either be integrated into the corresponding host genome or capable of infecting the host. Predicting regulatory relationships between mismatched phage-host pairs is meaningless.

Dependencies

• PSOSP is a Python script that relies on:

  DIAMOND=2.1.8
  MEME=5.5.5
  Python=3.12
  scikit-learn=1.6.1
  Prodigal=2.6.3
  biopython=1.85
  checkv=1.0.3

Installation

(1) conda (recommended, easiest way to install)

• install conda and add channels (If already installed, please skip)

  wget -c https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
  bash Miniconda3-latest-Linux-x86_64.sh
  source ~/.bashrc
  conda config --add channels bioconda
  conda config --add channels conda-forge

• install PSOSP

  conda create -n psosp psosp
  conda activate psosp

  test installation:psosp test usage: psosp -h

• If you need CheckV results, please download CheckV database

  checkv download_database ./

(2) git (install dependencies mentioned above first)

git clone https://github.com/mujiezhang/PSOSP.git
cd PSOSP
pip install -e .

test installation:psosp test

Input files

PSOSP needs two files as inputs,i.e.,

• -hf: a host genome in fasta format
• -vf: phage genome in fasta format

other parameters

• -wd: woking path to save result files
• -faa: host protein sequences in fasta format (optional)
• -db: checkv reference database path (optional)

How to run

The users can only specify the required parameters:

• install through conda

  psosp -hf /path/to/host-genome.fasta -vf /path/to/virus-genome.fasta -wd output_dir -db /path/to/checkv-db

• using example data in github or in zenodo for a test:

  psosp -hf test/data/host_wp2.fna -vf test/data/virus_wp2-phage-sp1-sp2-sp3.fna -wd test/test-result -db /path/to/checkv-db

Outputs

In this example, the results of PSOSP’s analysis will be written to the test/test-result directory, which will look like this:

test/test-result
├── virus_wp2-phage-sp1-sp2-sp3_checkv
├── host_wp2.fna.faa_lexa_blast.tsv
├── host_wp2.fna_whole_genome_HI.tsv
├── host_wp2_prodigal.faa
╰── virus_wp2-phage-sp1-sp2-sp3_prediction.tsv

1. virus_wp2-phage-sp1-sp2-sp3_checkv: this directory contain results of CheckV
2. host_wp2.fna.faa_lexa_blast.tsv: blast result of LexA protein
3. host_wp2.fna_whole_genome_HI.tsv: HI clusters of all potential SOS box in host genome using MeanShift method.
4. host_wp2_prodigal.faa: protein sequences produced by prodigal
5. virus_wp2-phage-sp1-sp2-sp3_prediction.tsv: prediction results of PSOSP

A detailed overview of virus_wp2-phage-sp1-sp2-sp3_prediction.tsv:

host	virus	prediction_result	prediction_quality	completeness	contamination	viral-HI(min)	box-seq	box-seq_start_pos	box-seq_strand	confidence_window_lower	confidence_window_upper	blast_status	fimo_status
host_wp2.fna	sp1	SiP (SOS-independent Prophage)	High	100.0	0.0	14.2017	CACTGTATTATTATACCACA	29652	-	11.8522	13.3752	Blast_OK	Fimo_OK
host_wp2.fna	sp2	SdP (SOS-dependent Prophage)	High	100.0	0.0	11.7239	TTATGTATGTATATTCAGCA	17757	-	11.8522	13.3752	Blast_OK	Fimo_OK
host_wp2.fna	sp3	SiP (SOS-independent Prophage)	High	99.62	0.0	15.4715	CACTGTATAAAAAAACATAC	1225	-	11.8522	13.3752	Blast_OK	Fimo_OK

• host: Input host filename
• virus: Phage identifier in input FASTA
• prediction_result: prediction induction mode of PSOSP. SiPs: SOS-independent Prophage; SuPs: SOS-uncertain Prophage; SdPs: SOS-dependent Prophage
• prediction quality: High for phage completeness between 90%-100% or for phages predicted as SdPs; Medium for phage completeness between 50%-90%; Low for phage completeness lower than 50%
• completeness: Estimated phage completeness (CheckV)
• contamination: Estimated phage contamination (CheckV)
• viral-HI(min): Minimal HI in phage genome
• box-seq: sequence of potential sos box with minimal HI
• box-seq_start_pos: box start position in phage genome
• box-seq_strand: + (forward) or - (reverse)
• confidence_window_lower: Threshold HI_c1 (HI_min ≤ HI_c1 → SdP)
• confidence_window_upper: Threshold HI_c2 (HI_min ≥ HI_c2 → SiP)
• blast_status: ‘Blast_OK’ (LexA homologs found) or ‘-‘ if absent
• fimo_status: ‘Fimo_OK’ (SOS-box detected upstream of LexA) or ‘-‘ if absent

Citation

Hao, Yali, Mujie Zhang, Xinjuan Lei, Chengrui Zhu, Taoliang Zhang, Yanping Zheng, Xiang Xiao, and Huahua Jian. 2025. “PSOSP Uncovers Pervasive SOS‐Independent Prophages With Distinct Genomic and Host Traits in Bacterial Genomes.” iMeta e70073. https://doi.org/10.1002/imt2.70073