FatePredictor: Cell fate decision-making prediction with an ensemble deep learning model

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Cell differentiation involves critical transitions known as cell fate decision-making or bifurcation.
• Understanding these bifurcations offers insights into fundamental biological mechanisms.
• Conventional methods struggle to accurately predict these transitions and infer dynamics from single-cell RNA sequencing (scRNA-seq) data.

Purpose of the Study:

• To develop a novel computational framework, FatePredictor, for predicting cell fate bifurcation.
• To accurately infer cell fate dynamics from single-cell data.
• To identify key genes and pathways involved in cellular processes.

Main Methods:

• FatePredictor integrates bifurcation theory and optimal transport theory.
• It employs dynamic unbalanced optimal transport to reconstruct cell trajectories.
• An ensemble deep learning model predicts the dynamics of cell fate bifurcation.

Main Results:

• FatePredictor accurately predicts bifurcations in simulated and real scRNA-seq data.
• The framework outperforms existing methods in predicting complex biological system dynamics.
• It successfully unveils intricate cellular trajectories and identifies key genes/pathways.

Conclusions:

• FatePredictor is a powerful and user-friendly tool for analyzing cell fate dynamics.
• It advances the prediction of critical transitions in biological systems.
• The framework provides deeper insights into the mechanisms governing cellular processes.