lncrnapi 0.1.5

A CLI tool for predicting lncRNA–Protein interactions using transformer embeddings and CatBoost

🧬 lncrna-PI - LncRNA–Protein Interaction Prediction

lncrnaPI is a command-line tool for predicting lncRNA–Protein interactions using pre-trained language models (DNABERT-2 and ESM-2) for sequence embedding and a CatBoost classifier for interaction probability estimation.

---

🚀 Overview

This standalone script enables large-scale prediction of interactions between lncRNA and protein sequences.
It leverages state-of-the-art transformer models to extract biologically meaningful embeddings and a pre-trained CatBoost model to compute interaction probabilities.

---

📦 Features

• Supports FASTA input for lncRNA and protein sequences.
• Generates embeddings using:
  • 🧬 DNABERT-2 (zhihan1996/DNABERT-2-117M) for lncRNAs
  • 🧫 ESM-2 (facebook/esm2_t30_150M_UR50D) for proteins
• Predicts interaction probabilities using a CatBoost classifier.
• Supports GPU acceleration (CUDA / MPS) for faster inference.
• Outputs results in CSV format.

---

🧰 Installation

Install the package using the following code from the command line:

pip install lncrnapi

---

⚙️ Usage

Run the script directly from the command line:

lncrnapi  --lncrna_fasta /path/to/lncrnas.fasta     --protein_fasta /path/to/proteins.fasta     --model_path /path/to/saved_model.joblib     --output_file /path/to/results.csv

Arguments

Argument	Description	Required
--lncrna_fasta	Path to the FASTA file containing lncRNA sequences.	✅
--protein_fasta	Path to the FASTA file containing protein sequences.	✅
--model_path	Path to the pre-trained CatBoost model file (.cbm, .joblib, or .pkl).	✅
--output_file	Path to save the CSV file with predicted probabilities.	✅

---

🧠 How It Works

1. Model Loading
   The tool loads the DNABERT-2 and ESM-2 models from Hugging Face.

2. FASTA Parsing
   Extracts sequence IDs and corresponding sequences from input FASTA files.

3. Embedding Generation
   Computes mean pooled embeddings for each sequence using transformer hidden states.

4. Prediction
   Concatenates embeddings (lncRNA + protein) and predicts the interaction probability using the CatBoost model.

5. Output
   Generates a .csv file containing:

   • LncRNA_ID
   • Protein_ID
   • Interaction_Probability

---

📊 Example Output

LncRNA_ID	Protein_ID	Interaction_Probability
lnc001	P12345	0.9421
lnc002	Q8N6T7	0.3175

---

⚡ Hardware Acceleration

The script automatically detects and uses available hardware:

• ✅ CUDA GPU (NVIDIA)
• ✅ MPS (Apple Silicon)
• ⚠️ CPU (fallback)

---

🧩 Model Formats Supported

Format	Description
.cbm	Native CatBoost model format
.joblib	Joblib-serialized model
.pkl	Pickle-based serialized model

---

🛠 Troubleshooting

Issue	Possible Cause	Solution
Model file not found	Wrong --model_path	Check the file path
No sequences found in FASTA	Invalid FASTA format	Ensure > headers are present
safetensors error	Missing library	Install with pip install safetensors
Slow performance	CPU usage	Use GPU-enabled environment

---

📁 Output Example

$ head results.csv
LncRNA_ID,Protein_ID,Interaction_Probability
lnc001,P12345,0.9421
lnc002,Q8N6T7,0.3175
lnc003,O76074,0.7814

---

📜 Citation

If you use this tool in your research, please cite:

Your Name et al.
A Deep Learning Framework for lncRNA–Protein Interaction Prediction Using Transformer-Based Sequence Embeddings (2025)

---

🧩 Repository Structure

├── predict_interaction.py       # Main CLI script
├── README.md                    # Documentation
└── example/
    ├── lncrnas.fasta
    ├── proteins.fasta
    └── saved_model.joblib

---