École Nationale Supérieure de
Techniques Avancées

Abstract

In this project, we undertook a comprehensive analysis of the dynamics of the CCR5 receptor within the cell membrane, focusing on the precise detection of movement change points throughout its trajectories. The receptor’s movements were modeled using several stochastic processes. These models allowed us to simulate various dynamic behaviors of the receptor. To characterize these movements, we employed advanced analytical techniques such as mean squared displacement (MSD) and directional persistence. These methods were integrated with machine learning algorithms, notably Random Forest, to classify the types of movements. Additionally, a carefully calibrated neural network model, applied with a sliding window, was used to provide confidence scores. This approach not only enabled us to classify the types of CCR5 receptor movements but also to pinpoint the change points in its trajectories. Through this integration of analytical methods and machine learning, we achieved a detailed understanding of the complex dynamics of the CCR5 receptor, enhancing our comprehension of its interactions and behavior within the cell membrane.