DescriptionIn this thesis, we develop data driven formalism for analysing features that govern dynamics in biological systems. We first consider the social amoeba Dictyostelium discoideum as a model organism. D. discoideum cells form protrusions and migrate via cytoskeletal reorganization driven by coordinated waves of actin polymerization and depolymerization. Assembly and disassembly of actin filaments are regulated by a complex network of biochemical reactions, exhibiting sensitivity to external physical cues such as stiffness, composition, and surface topography of the extracellular matrix. The dynamics and topology of actin waves in moving D. discoideum cells are affected by the nanotopography of the cell surface and the presence of the external electric field. In the first part of this study, we employ machine learning techniques to predict the type of the extracellular environment by observing actin waves either frame-by-frame or over a period of time, and identify key visual features that help classify cell motion by the microenvironment type. In the second part, we focus on learning dynamical rules with the goal of predicting dynamics from experimental observations of signaling in biological systems. In recent years, Dynamic Mode Decomposition (DMD), along with Koopman Operator Theory, has been used as a data-driven technique to understand complex dynamics in high-dimensional systems. We reformulate DMD using kernel methods and provide a formalism to estimate Koopman operator from observed data, which can be used to predict temporal evolution of dynamical systems. We showcase some applications of our formalism on both synthetic and real datasets. We expect our computational approach to be useful in many settings where non-trivial collective dynamics is observed.