pratik0297/antifp2 repository overview

⁠AntiFP2: Antifungal Protein Prediction Toolkit (Docker Version)

This repository provides a fully containerized version of AntiFP2, a toolkit for antifungal protein prediction using:

• ESM2 (fine-tuned language model)
• Machine Learning (PAAC features)
• BLAST (sequence similarity adjustment)
• MERCI (motif adjustment)
• Prokka (for genome annotation in meta-pipelines)

All dependencies (Python, ESM2 models, BLAST, MERCI, Prokka, etc.) are pre-installed inside the Docker image.

⁠🚀 Running the Docker Container

Instead of running individual scripts directly, you should always enter the Docker container first.

⁠Start an interactive shell inside the container

docker run -it --rm antifp2 bash

By default, this gives you a working shell inside the container.

If you want to work on files stored on your host machine, mount them into the container:

docker run -it --rm -v /absolute/path/to/data:/workspace -w /workspace antifp2 bash

• /absolute/path/to/data → path on your host with FASTA/contigs files.
• /workspace → working directory inside the container (you will be placed here after startup).

⁠🧭 Master Launcher: antifp2.py

All methods are launched using a single master script:

python3 antifp2.py --method <method-name> [method-specific options]

⁠List all available methods

python3 antifp2.py --list

This will display:

• esm2
• esm2-hybrid
• esm2-meta
• ml-hybrid
• ml-hybrid-meta

Each corresponds to a backend script or combination pipeline.

⁠📖 Detailed Usage per Method

⁠1️⃣ ESM2 Only — esm2

Predict antifungal proteins using the fine-tuned ESM2 model (sequence-only, one-by-one).

⁠Arguments:

⁠Example:

python3 antifp2.py --method esm2 --input proteins.fasta --output esm2_predictions.csv --threshold 0.5

⁠2️⃣ ESM2 + BLAST + MERCI — esm2-hybrid

Combines ESM2 predictions with BLAST adjustments and motif analysis (MERCI).

⁠Arguments:

⁠Example:

python3 antifp2.py --method esm2-hybrid --input proteins.fasta --outdir esm2_hybrid_results --threshold 0.6

⁠3️⃣ MetaPipeline (Prokka + ESM2) — esm2-meta

Runs Prokka genome annotation first, then applies ESM2 predictions on predicted proteins.

⁠Arguments:

⁠Example:

python3 antifp2.py --method esm2-meta --input contigs.fasta --outdir esm2_meta_out --threads 8

⁠4️⃣ ML Hybrid (PAAC + BLAST + MERCI) — ml-hybrid

Uses ML classifier (PAAC features) combined with BLAST and MERCI adjustments.

⁠Arguments:

⁠Example:

python3 antifp2.py --method ml-hybrid --input proteins.fasta --outdir ml_hybrid_out --threshold 0.5

⁠5️⃣ MetaPipeline (Prokka + ML Hybrid) — ml-hybrid-meta

Runs Prokka annotation, then applies ML + BLAST + MERCI hybrid classifier.

⁠Arguments:

⁠Example:

python3 antifp2.py --method ml-hybrid-meta --input contigs.fasta --outdir ml_hybrid_meta_out --threads 8 

⁠📂 Outputs

Depending on the method, outputs include:

⁠🧪 Example Full Workflow

Assume your FASTA files are in /home/user/antifp2_data.

1. Start the container and mount your data:

docker run -it --rm -v /home/user/antifp2_data:/workspace -w /workspace antifp2 bash

1. Inside the container, run one of the pipelines. Example with ML hybrid:

python3 antifp2.py --method ml-hybrid --input proteins.fasta --outdir results

1. Outputs (results/) will appear both inside the container and in your host directory (/home/user/antifp2_data/results).

⁠🧹 Cleanup

Use --no-cleanup if you want to keep intermediate .faa, .locate, .csv, or Prokka files for debugging. Otherwise, they are automatically removed.

⁠🍏 Running AntiFP2 on macOS (Apple Silicon)

AntiFP2 runs inside a Linux-based Docker container, which is fully compatible with Intel-based Macs and Apple Silicon (M1/M2/M3) using emulation.

⁠✅ Step-by-Step Setup for Apple Silicon (M1/M2/M3)

1. Install Docker Desktop for macOS

   • Download from: https://www.docker.com/products/docker-desktop⁠
   • During install, enable Rosetta emulation support when prompted (or install it manually using the command below).

2. Enable Rosetta Emulation (only needed once):

softwareupdate --install-rosetta

1. Pull and run AntiFP2 with x86_64 compatibility:

docker run --platform linux/amd64 --rm -it -v /absolute/path/on/mac:/data pratik0297/antifp2

1. Run the pipeline inside the container as usual, for example:

python3 antifp2.py --method

esm2-meta --input /data/your_contigs.fasta --outdir /data/esm2_meta --threads 4 --metagenome

### 🧪 macOS Example

Assuming your FASTA file is at `~/Downloads/testing/first_100.fasta`:

```bash
docker run --platform linux/amd64 --rm -it -v ~/Downloads/testing:/data pratik0297/antifp2

Then, inside the container:

python3 antifp2.py --method esm2-meta --input /data/test_contigs.fasta --outdir /data/esm2_meta

⁠🧯 Troubleshooting

• If you get Bad CPU type in executable or Illegal instruction errors, it likely means:

  • You didn't use --platform linux/amd64
  • Rosetta 2 isn’t installed

• Prokka and BLAST inside Docker will work even if they don't install on native macOS.

⁠✅ Citation

If you use AntiFP2 in your research, please cite: (Add citation / preprint link here)

⁠👨‍💻 Support

For issues, please:

• Open an issue on the GitHub repository https://github.com/raghavagps/antifp2⁠, or
• Contact the developer directly.[email protected]⁠