Laboratory of the Physics of Biological Systems (LPBS) - C. elegans Movement Analysis

Project Overview

This project implements a comprehensive machine learning pipeline for analyzing movement patterns of Caenorhabditis elegans worms tracked on plates. The system can classify worms that have been administered drugs versus control worms and predict death proximity.

📖 For detailed methodology, architecture decisions, data leakage prevention strategies, and main results see METHODOLOGY.md or REPORT.md.

Quick Start

1. Environment Setup

git clone https://github.com/lysandre-c/LPBS-Celegans.git
cd LPBS
pip install -r requirements.txt

2. Data Processing Pipeline

# 1. Preprocess raw data (splits into segments, normalizes)
python preprocessing.py

# 2. Extract features from preprocessed trajectories
python feature_extraction.py

3. Model Training & Evaluation

Feature-Based Classification

# Train feature-based models (Random Forest, Gradient Boosting, MLP)
python feature_segment_classification.py

Time Series Classification

# Train time series models (CNN, LSTM)
python ts_segment_classification.py

Death Proximity Prediction

# Classify segments as "close to death" or not (binary classification)
python death_proximity_predictor.py

# Predict number of segments remaining until death (regression)
python death_proximity_regressor.py