Machine learning inference of continuous single-cell state transitions during myoblast differentiation and fusion

Abstract

(Figure presented.) The prediction certainty of machine learning classification models can be used as a continuous measurement to quantitatively monitor single cell state transitions, as demonstrated for myoblast differentiation during muscle fiber formation. Live imaged single myoblast continuous differentiation states are computationally derived from motility and actin dynamics. The model distinguishes between cells that differentiated but failed to fuse to predict molecules specifically involved in fusion, as well as changes in actin dynamics. Mass spectrometry supports these in silico predictions and suggests novel fusion and maturation regulators downstream of differentiation. p38 is essential for the transition from terminal differentiation to fusion.