AmyloDeep

AmyloDeep: pLM-based ensemble model for predicting amyloid propensity from the amino acid sequence

AmyloDeep is a Python package that uses ensemble model to predict amyloidogenic regions in protein sequences using a rolling window approach.

Features

• Multi-model ensemble: Combines 5 different models for robust predictions
• Rolling window analysis: Analyzes sequences using sliding windows of configurable size
• Easy-to-use API: Simple Python interface and command-line tool
• Web interface: Light version of tool is available at amylodeep.com

Installation

From PyPI (recommended)

pip install amylodeep

From source

git clone https://github.com/AlisaDavtyan/protein_classification.git
cd amylodeep
pip install amylodeep

Quick Start

from amylodeep import predict_ensemble_rolling

# Predict amyloid propensity for a protein sequence
sequence = "MKTFFFLLLLFTIGFCYVQFSKLKLENLHFKDNSEGLKNGGLQRQLGLTLKFNSNSLHHTSNL"
window_size = 10
result = predict_ensemble_rolling(sequence, window_size = window_size)

print(f"Average probability: {result['avg_probability']:.4f}")
print(f"Maximum probability: {result['max_probability']:.4f}")

# Access position-wise probabilities
for position, probability in result['position_probs']:
    print(f"Position {position}: {probability:.4f}")
  
# Plot probabilities
import numpy as np
import matplotlib.pyplot as plt
positions, probs = zip(*result['position_probs'])
x = np.arange(0, len(sequence) - window_size + 1)
bar_colors = [
                (0, 0, 1, 0.8) if p > 0.8 else (0, 0, 1, 0.6) if p > 0.5 else (0, 0, 1, 0.2) for p in probs
            ]
fig, ax = plt.subplots(figsize=(10, 5))
ax.bar(x, probs, color=bar_colors, width=1, edgecolor="black")
ax.set_ylabel("Probability", fontsize=12)
ax.set_xlabel("Residue", fontsize=12)
L = len(sequence)

ax.set_xlim(-1, L - window_size + 1)

if L < 100:
   ax.set_xticks(np.arange(0, L+1, 5))
else:
  step = int(np.ceil(L/5/10) * 10)
  tick_labels = np.arange(0, L+1, step)
  tick_positions = np.minimum(tick_labels, L - window_size)
  ax.set_xticks(tick_positions)
  ax.set_xticklabels([str(t) for t in tick_labels])

ax.axhline(y=0.5, color='green', linestyle='--', alpha=0.7)
ax.axhline(y=0.8, color='red', linestyle='--', alpha=0.7)
ax.tick_params(axis='both', labelsize=12)
ax.set_title('Amyloidogenicity probability per window ', fontsize=16)

Command Line Interface

# Basic prediction
amylodeep "MKTFFFLLLLFTIGFCYVQFSKLKLENLHFKDNSEGLKNGGLQRQLGLTLKFNSNSLHHTSNL"

# With custom window size
amylodeep "SEQUENCE" --window-size 10

# Save results to file
amylodeep "SEQUENCE" --output results.json --format json

# CSV output
amylodeep "SEQUENCE" --output results.csv --format csv

Requirements

• torch>=1.12.0
• transformers>=4.30.0
• huggingface_hub>=0.14.0
• xgboost>=1.7.0
• numpy>=1.20
• pandas>=1.3
• scikit-learn>=1.0
• jax-unirep>=2.0.0

Main Functions

predict_ensemble_rolling(sequence, window_size=10)

Predict amyloid propensity for a protein sequence using rolling window analysis.

Parameters:

• sequence (str): Protein sequence (amino acid letters)
• window_size (int): Size of the rolling window (default: 10)

Returns: Dictionary containing:

• position_probs: List of (position, probability) tuples
• avg_probability: Average probability across all windows
• max_probability: Maximum probability across all windows
• sequence_length: Length of the input sequence
• num_windows: Number of windows analyzed

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If you use AmyloDeep in your research, please cite:

@software{amylodeep2025,
  title={AmyloDeep: pLM-based ensemble model for predicting amyloid propensity from the amino acid sequence},
  author={Alisa Davtyan, Anahit Khachatryan, Rafayel Petrosyan},
  year={2025},
  url={https://github.com/AlisaDavtyan/protein_classification}
}

Support

For questions and support:

• Open an issue on GitHub
• Contact: alisadavtyan7@gmail.com , xachatryan96an@gmail.com