Classification of Single Molecular Motor Events

Abstract

This paper describes the development of two methods for automatic classification of molecular events. The methods are applied to experimental results obtained to study the interaction of a single microtubule with a single Ncd molecular motor. The data were previously obtained using optical trap assay, and have a very low signal-to-noise ratio of approximately 0.1. The first method can be viewed as a syntactic classification. In an alternative approach, radial basis neural networks, trained with simulated data generated by a system of stochastic, Langevin-like differential equations, are used for classification. Following classification, the molecular events are ensemble averaged separately for different types of events. The analysis of the results suggests the existence of Ncd-microtubule interaction events, which deviate from the traditional view on the kinetics of the Ncd-microtubule interactions. This paper employs a novel paradigm, which emphasizes the importance of the model-based, time-resolved filtering of the experimental data on molecular interactions, especially for the case of biological molecules, followed by the analysis of different groups of events, automatically segregated according to features revealed during the single realization analysis of the experimental results.