IMMUNOLOGY2026™ Conference Recordings For Attendees-Uncovering cellular programs and regulatory circuits governing cell fate bifurcations using interpretable machine learning

Uncovering cellular programs and regulatory circuits governing cell fate bifurcations using interpretable machine learning

Back to course

Video Transcription

Video Summary

Zarifeh Heydari presents FIREFATE, a conceptual framework for decoding cell fate decisions by combining interpretable machine learning with mechanistic modeling. FIREFATE aims to identify how fate choices are encoded in regulatory programs, infer dynamic TF activity, and even predict fate before commitment. The approach was applied to two immune systems: B-cell differentiation and T-cell exhaustion. In B cells, the team built high-resolution, state-specific gene regulatory networks from single-cell multiome data, then used the SLIDE machine-learning model to identify latent gene programs. By overlaying these latent factors onto the networks, they prioritized core sub-networks and key TFs such as PAX5 and PRDM1. These predictions matched perturbation experiments and benchmarked well against orthogonal data. Using dynamic models, they also uncovered episodic TF activity and fate predisposition in early cells. FIREFATE is proposed as a generalizable strategy for studying diverse cell fate systems.

Meta Tag

Date April 19, 2026 8:52 AM - 9:05 AM

Room 102

Session Development of Statistical, Machine Learning, and Genetic Models

Speaker Zarifeh Heidari Rarani

Track Computational and Systems Immunology (COMP)

Year 2026

Keywords

FIREFATE

cell fate decisions

gene regulatory networks

B-cell differentiation

T-cell exhaustion

April 19, 2026 8:52 AM - 9:05 AM

102

Development of Statistical, Machine Learning, and Genetic Models

Zarifeh Heidari Rarani

Computational and Systems Immunology (COMP)

2026