OmniEff - Effector Prediction Pipeline

OmniEff is a comprehensive Snakemake-based pipeline for predicting effector proteins from genomic assemblies. The pipeline integrates multiple prediction tools including SignalP, EffectorP, DeepRedEff, and others to provide a comprehensive analysis of potential effector proteins.

Quick Start (TSL HPC - Installed Version)

If OmniEff is already installed on your system, you can start using it immediately:

# 1. Load the package
source package omnieff-0.0.2

# 2. Create a working directory and assemblies directory
mkdir my_effector_analysis
cd my_effector_analysis
mkdir assemblies

# 3. Add your protein FASTA files to assemblies directory
cp your_proteins.faa assemblies/assembly_name.faa

# 4. Copy the configuration template
cp $OMNIEFF_INSTALL_PREFIX/lib/config.yaml.template config.yaml

# 5. Edit config.yaml with your settings
nano config.yaml
# Required: scratch, workdir, wckey
# Optional: assemblies_dir (defaults to ./assemblies)

# 6. Run the pipeline
omnieff_run --config config.yaml

Important: Input files must be named {assembly_name}.faa (see Input File Preparation below).

For system administrators wanting to install OmniEff, please see the install instructions in the source repo INSTALL.md.

---

Overview

The pipeline performs the following analyses:

• Signal peptide prediction: SignalP3, SignalP4, SignalP5, SignalP6, Phobius
• Subcellular localization: TargetP, TMHMM, LOCALIZER, ApoplastP
• Effector prediction: EffectorP1, EffectorP2, EffectorP3 (fungal/non-fungal), DeepRedEff (fungi/oomycete)
• Domain analysis: PFAM domain scanning
• Scoring and filtering: Custom scoring algorithm to identify high-confidence effectors

Installation and Setup

Prerequisites

1. Snakemake (version 9.3.0 or higher)
2. Singularity for container support
3. SLURM job scheduler
4. Python 3.12 with required packages

Setup

1. Clone the repository:

git clone TeamMacLean/omnieff
cd omnieff

2. Create the conda environment:

mamba env create -p ./snakemake_env -f lib/snakemake_env.yaml

3. Activate the environment:

source activate snakemake_env

Configuration

Input File Preparation

CRITICAL: The pipeline expects input files to be named with a specific pattern in the assemblies directory.

File Naming Convention

Your input protein FASTA files must be placed in the assemblies directory with the following naming pattern:

{assembly_name}.faa

Examples:

• canu.faa
• flye.faa
• spades.faa

NOT:

• canu_proteins.faa ❌
• canu_proteins.fasta ❌
• canu.fasta ❌

The get_assemblies() function in the workflow automatically detects assembly names by removing the .faa extension from files in the assemblies directory.

Configuration File

Edit lib/config.yaml to set your parameters:

# Working directory settings
scratch: "/tsl/scratch/your_username"
workdir: "your_project_name"

# Input files directory
assemblies_dir: "./assemblies"

# Container settings
singularity_image: "/tsl/data/singularity_images/predector/predector.sif"

# SLURM settings
wckey: "your_wckey"
main_job_partition: "tsl-medium"

# Profile directory for Snakemake
profile_dir: "./profiles"

Key Configuration Parameters

• scratch: Base directory for all outputs
• workdir: Project-specific subdirectory
• assemblies_dir: Directory containing input .faa files (relative to where you run the command, or absolute path)
• singularity_image: Path to the Singularity container - you can leave this as the default
• wckey: SLURM wckey for job accounting
• main_job_partition: SLURM partition for job submission
• profile_dir: Directory containing Snakemake profiles

Usage

Installed Version (Recommended)

If using the installed version on TSL HPC:

# Load the package
source package omnieff-0.0.1

# Run the pipeline
omnieff_run --config config.yaml

# Dry run (test without executing)
omnieff_run --dry-run --config config.yaml

# View version
omnieff_run --version

Development Version (From Git Repository)

If running from a cloned repository:

# Run the complete pipeline
python src/run_workflow.py --config lib/config.yaml

# Dry run (test without executing)
python src/run_workflow.py --dry-run

Common Options

The following options work with both omnieff_run (installed) and python src/run_workflow.py (development):

Dry run (test without executing):

omnieff_run --dry-run --config config.yaml
# OR
python src/run_workflow.py --dry-run

Run until a specific rule:

python src/run_workflow.py --rule score_proteins

Generate DAG visualization:

python src/run_workflow.py --dag

Unlock a locked directory:

python src/run_workflow.py --unlock

Debug mode:

python src/run_workflow.py --debug

Force run a rule:

python src/run_workflow.py --rule score_proteins --force

Output Structure

The pipeline generates outputs in the following structure:

{scratch}/{workdir}/
├── assemblies/           # Input FASTA files
├── signalp3/            # SignalP3 results
├── signalp4/            # SignalP4 results
├── signalp5/            # SignalP5 results
├── signalp6/            # SignalP6 results
├── phobius/             # Phobius results
├── targetp/             # TargetP results
├── tmhmm/               # TMHMM results
├── effectorp1/          # EffectorP1 results
├── effectorp2/          # EffectorP2 results
├── effectorp3/          # EffectorP3 results
├── localizer/           # LOCALIZER results
├── apoplastp/           # ApoplastP results
├── deepredeff/          # DeepRedEff results
├── pfam/                # PFAM domain results
├── parsed/              # Parsed tool outputs
├── scored/              # Scored proteins
├── filtered/            # Final filtered effectors
└── logs/                # Log files

Key Output Files

Parsed Results (parsed/{assembly}_tool_results.csv): Combined results from all prediction tools for each protein. Contains columns for each tool's predictions and scores.

Scored Results (scored/{assembly}_scored.csv): Parsed results with added custom_score and score_rank columns based on the scoring algorithm.

Filtered Results (filtered/{assembly}_filtered.csv): High-confidence effector candidates that passed filtering thresholds.

Filtered Sequences (filtered/{assembly}_filtered.fasta): FASTA file containing only the filtered effector protein sequences.

Understanding Output Columns

RESULTS_MISSING: This value appears in output files when a particular analysis was not performed or could not be completed. For example:

• signalp3_nn and signalp3_nn_d are always RESULTS_MISSING because the pipeline only runs SignalP3 HMM method
• Other fields may show RESULTS_MISSING if a tool failed or if the protein was not analyzed by that tool

NA: Used for numeric fields where the value is not applicable or not available (e.g., EffectorP3 probabilities when a protein is predicted as non-effector).

Workflow Rules

Core Analysis Rules

Note on SignalP3: The pipeline runs only SignalP3 HMM method. The Neural Network (NN) method is not executed, so signalp3_nn fields will show RESULTS_MISSING in the output files. This is expected behavior.

• run_signalp3_hmm: Signal peptide prediction (HMM method only)
• run_signalp4: Signal peptide prediction (v4)
• run_signalp5: Signal peptide prediction (v5)
• run_signalp6: Signal peptide prediction (v6)
• run_phobius: Transmembrane topology prediction
• run_targetp: Subcellular localization prediction
• run_tmhmm: Transmembrane helix prediction
• run_effectorp1: Effector prediction (v1)
• run_effectorp2: Effector prediction (v2)
• run_effectorp3_fungal: Effector prediction (v3, fungal)
• run_effectorp3_nonfungal: Effector prediction (v3, non-fungal)
• run_localizer: Subcellular localization
• run_apoplastp: Apoplast localization prediction
• run_deepredeff_fungi: Deep learning effector prediction (fungi)
• run_deepredeff_oomycete: Deep learning effector prediction (oomycete)
• run_pfam_scan: PFAM domain scanning

Processing Rules

• parse_tool_outputs: Parse and combine all tool outputs
• score_proteins: Apply scoring algorithm
• filter_proteins: Filter high-confidence effectors
• extract_filtered_sequences: Extract final effector sequences

Troubleshooting

Common Issues

1. Missing input files: Ensure your FASTA files are named correctly ({assembly}.faa)
2. Empty assemblies list: Check that the assemblies directory exists and contains properly named files
3. SLURM job failures: Verify your SLURM configuration and wckey settings
4. Container issues: Ensure the Singularity container path is correct

Debug Mode

Use debug mode to get detailed information:

python src/run_workflow.py --debug

This will provide verbose output and help identify issues in the workflow.

Dependencies

The pipeline uses the Predector Singularity container which includes:

• SignalP (v3, v4, v5, v6)
• EffectorP (v1, v2, v3)
• DeepRedEff
• LOCALIZER
• ApoplastP
• TargetP
• TMHMM
• Phobius
• HMMER (for PFAM scanning)

Citation

If you use this pipeline in your research, please cite the original tools and this pipeline.

License

Internal use only

Support

For issues and questions, please open an issue on the repository or contact the TSL bioinformatics team.