algpred3 1.0

Prediction of allergenic and non-allergenic peptides

AlgPred3

A computational framework for predicting and designing allergenic and non-allergenic peptides using machine learning and composition-based features.

---

📌 Introduction

AlgPred3 is developed to identify allergenic potential of peptides based on their primary sequence. It integrates feature-based machine learning approaches with curated allergen datasets to provide robust predictions. 🔗 Visit the web server for more information: Algpred3

The tool supports:

• Prediction of allergenic peptides
• Protein scanning for allergenic regions
• Design of peptide mutants

It uses compositional and physicochemical descriptors such as:

• AAC (Amino Acid Composition)
• DPC (Dipeptide Composition)
• PAAC, APAAC
• CeTD, DDR, BTC

---

📚 Reference

AlgPred3 Raghava Group, IIIT-Delhi

---

🧪 Quick Start for Reproducibility

# 1. Clone the repository
git clone https://github.com/raghavagps/algpred3.git
cd algpred3
# 2. Set up the environment (conda recommended)
conda env create -f environment.yml
conda activate algpred3

# 3. Run help
python algpred3.py -h

# 4. Run prediction
python algpred3.py -i example.fasta -o output.csv -j pred

---

🛠️ Installation Options

🧰 Pip Installation

pip install algpred3

Check options:

algpred3 -h

---

🔹 Standalone Installation

AlgPred3 is written in Python 3 and requires:

✅ Required Libraries

python=3.8+

📦 Install Dependencies

pip install scikit-learn
pip install pandas
pip install joblib

---

🔹 Automatic Dependency Setup

AlgPred3 automatically downloads required files during runtime:

• Model file (algpred3_model.sav)
• Feature column file (columns.csv)
• pfeature binary

This behavior is implemented directly in the script.

---

⚠️ Important Notes

• First run requires internet connection for auto-download.
• Input sequences must contain only valid amino acids: ACDEFGHIKLMNPQRSTVWY
• Invalid sequences are removed and logged in:

stand_error.log

---

🔬 Classification

AlgPred3 classifies peptides into:

• Allergen
• Non-Allergen

based on machine learning predictions using extracted sequence features.

---

🚀 Usage

🔹 Minimum Usage

python algpred3.py -h

---

🔹 Full Usage

usage: algpred3.py [-h]
                  -i INPUT
                  [-o OUTPUT]
                  -j {pred,scan,des}
                  [-t THRESHOLD]
                  [-l LENGTH]
                  [-s STEP]

---

📌 Arguments

Argument	Description
-i INPUT	Input FASTA or sequence file
-o OUTPUT	Output CSV file (default: final_predictions.csv)
-j	Job type: pred, scan, des
-t	Threshold (default: 0.5)
-l	Window length (scan mode only)
-s	Step size (default: 1)

---

📂 Input & Output Files

✅ Input Formats

1. FASTA format
2. Plain text format (one sequence per line)

---

✅ Output File

• Results are saved in CSV format

• Includes:

  • Sequence ID
  • Prediction score
  • Allergen / Non-Allergen status

---

🔍 Jobs & Features

🔹 Job Types

Job	Description
🔹 pred	Predict allergenic potential of sequences
🔹 scan	Identify allergenic regions in proteins
🔹 des	Generate and evaluate all possible mutants

---

🔹 Scan Mode

• Uses sliding window approach
• Generates peptide fragments
• Predicts allergenic regions

---

🔹 Design Mode

• Generates all single amino acid mutants

• Predicts allergenicity of each mutant

• Useful for:

  • Peptide optimization
  • Allergenicity reduction

---

⚙️ Workflow

1. Input sequence validation
2. Cleaning invalid sequences
3. Feature extraction (pfeature)
4. Feature selection
5. Model prediction
6. Output generation

---

📑 Package Contents

File	Description
algpred3.py	Main prediction script
columns.csv	Feature selection file
algpred3_model.sav	Trained ML model
pfeature_comp	Feature extraction binary
example.fasta	Sample input

---

📦 PIP Installation (Again for Reference)

pip install algpred3

---

🚀 Start predicting allergenic peptides with AlgPred3 today! 🔗 Visit: Algpred3