A method for predicting the neurotoxic activity of peptides and proteins.
NTxPred2 is designed to assist researchers in therapeutic peptide and protein development by providing advanced methods for quantifying and classifying neurotoxic peptides and proteins that target the central nervous system.
It employs large language model word embeddings as features for predicting neurotoxic activity. The final model offers Prediction, Protein-Scanning, and Design modules, implemented using machine learning and protein language models.
🔗 Visit the web server for more information: NTxPred2 Web Server
📖 Please cite relevant content for complete details, including the algorithm behind the approach.
Rathore AS, Jain S, Choudhury S, Raghava GPS. A large language model for predicting neurotoxic peptides and neurotoxins. Protein Sci. 2025 Aug;34(8):e70200.
Follow these steps to replicate the core results of our paper:
# 1. Clone the repository
git clone https://github.com/raghavagps/ntxpred2.git
cd ntxpred2
# 2. Set up the environment (conda recommended)
conda env create -f environment.yml
conda activate NTxPred2
# 3. Download pre-trained models
# Visit: https://webs.iiitd.edu.in/raghava/ntxpred2/download.html
# Download the model ZIP and extract it in the root directory
# 4. Run prediction on sample input
python ntxpred2.py -i example.fasta -o output.csv -m 1 -j 1 -wd working_direcotory_path
To install NTxPred2 via PIP, run:
pip install ntxpred2To check available options, type:
ntxpred2 -hNTxPred2 is written in Python 3 and requires the following dependencies:
python=3.10.7
pytorchAdditional required packages:
pip install scikit-learn==1.5.2
pip install pandas==1.5.3
pip install numpy==1.25.2
pip install torch==2.1.0
pip install transformers==4.34.0
pip install joblib==1.4.2
pip install onnxruntime==1.15.1
Bio (Biopython): 1.81
tqdm: 4.64.1
torch: 2.6.0- Create a new Conda environment:
conda env create -f environment.yml- Activate the environment:
conda activate NTxPred2- Due to the large size of the model file, the model directory has been compressed and uploaded.
- Download the zip file from Download Page.
- Extract the file before using the code or model.
NTxPred2 classifies peptides and proteins as neurotoxic or non-neurotoxic based on their primary sequence.
🔹 Model Options
- ESM2-t30 (Peptide Model): For sequences 7-50 amino acids long.
- ET (Protein Model): For sequences ≥ 51 amino acids.
- ET (Combined Model): For sequences of mixed length.
- Default Model: ESM2-t30 (Peptide Model), selected for best performance and efficiency.
ntxpred2.py -hTo run an example:
ntxpred2.py -i example.fastausage: ntxpred2.py [-h]
[-i INPUT]
[-o OUTPUT]
[-t THRESHOLD]
[-j {1,2,3,4}]
[-m {1,2,3}]
[-d {1,2}]
[-wd WORKING DIRECTORY]| Argument | Description |
|---|---|
-i INPUT |
Input: Peptide or protein sequence (FASTA format or simple format) |
-o OUTPUT |
Output file (default: outfile.csv) |
-t THRESHOLD |
Threshold (0-1, default: 0.5) |
-j {1,2,3,4} |
Job type: 1-Prediction, 2-Protein Scanning, 3-Design, 4-Design all possible mutants |
-m {1,2,3} |
Model selection: 1-ESM2-t30 (Peptides), 2-ET (Proteins), 3-ET (Combined) |
-wd WORKING |
Working directory for saving results |
NTxPred2 supports two formats:
- FASTA Format: (Example:
example.fasta) - Simple Format: (Example:
example.seq, each sequence on a new line)
- Results are saved in CSV format.
- If no output file is specified, results are stored in
outfile.csv.
| Job | Description |
|---|---|
| 1️⃣ Prediction | Predicts whether input peptide/protein is neurotoxic or not. |
| 2️⃣ Protein Scanning | Identifies neurotoxic regions in a protein sequence. |
| 3️⃣ Design | Generates mutant peptides/proteins with a single amino acid/dipeptide at a specified position. |
| 4️⃣ Design All Possible Mutants | Generates and predicts all possible mutants. |
| Option | Description |
|---|---|
-p POSITION |
Position to insert mutation (1-indexed) |
-r RESIDUES |
Mutated residues (single/double letter amino acid codes) |
-w {8-20} |
Window length (Protein Scan mode only, default: 12) |
-d {1,2} |
Display: 1-Neurotoxic only, 2-All peptides (default) |
| File | Description |
|---|---|
| INSTALLATION | Installation instructions |
| LICENSE | License information |
| README.md | This file |
| ntxpred2.py | Python program for classification |
| example.fasta | Example file (FASTA format) |
pip install ntxpred2Check options:
ntxpred2 -h🚀 Start predicting neurotoxicity with NTxPred2 today!
🔗 Visit: NTxPred2 Web Server