Interpretable machine learning coupled to gene regulatory networks uncovers subcircuits underlying cell fate decisions
- 1University of Pittsburgh
Abstract
Gene regulatory networks (GRNs) model causal linkages that control cell fate decisions and differentiation transitions. Prioritizing regulatory subnetworks underlying cell state differences is of critical importance, but current methods including those reliant on topological metrics introduce circularity as the metrics prioritizing TFs are computed from the same networks whose assumptions they inherit. Separately, interpretable machine learning methods can identify latent factors (LFs) that discriminate cellular states with formal statistical guarantees but do not model regulatory linkages. Here, we present FOCAL (Factor-Outcome Coupling for Assessment of Linkages), a paradigm to prioritize regulatory subnetworks by coupling state-specific and dynamic GRNs with outcome-supervised LFs learned using interpretable machine learning without reference to network topology. This shifts GRN focus from macroscopic TF nodes to state-specific and dynamic TF-gene linkages. In B and T cells, FOCAL identified GIFs (GRNs coupled to Interpretable latent Factors), prioritized regulatory subnetworks underlying established states as well as transient regulatory episodes preceding them. By coupling LFs learnt from perturbation experiments of lineage-defining TFs, FOCAL identified transcriptional predisposition to alternative fates within progenitor cell populations before overt differentiation. This uncovered a novel NFATC2-IRF8 interplay in activated B cells, that was validated by in-vitro and in-vivo genetic perturbations. The two transcription factors act cooperatively to restrain extrafollicular plasmablast differentiation and promote germinal center B cell fate.
Related
- High-dimensional population codes reveal interpretable and diverse features underlying visual perception
- Transcriptional regulatory programs underlying cellular and regional specialization in the late-gestation porcine brain
- Comprehensive in silico analysis reveals candidate regulatory mechanisms underlying selective cerebellar vulnerability in pontocerebellar hypoplasia
- Integrated Transcriptomic and Metabolomic Analyses Reveal Leaf Regulatory Mechanisms Underlying Low-Nitrogen Tolerance in Oat (Avena sativa L.)
- Gene regulatory logic underlying the differentiation of mixed lineages in human thymic epithelial cells
BibTeX
@misc{rarani2026interpretable,
title = {{Interpretable machine learning coupled to gene regulatory networks uncovers subcircuits underlying cell fate decisions}},
author = {Z. H. Rarani and S. Keshari and A. Sachan and A. Saini and N. A. Pease and J. Fan and A. Fu and Y. C. Liu and A. U. Gurkar and G. M. Delgoffe and H. Singh and J. Das},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.28.754797},
url = {https://www.biorxiv.org/content/10.64898/2026.09.28.754797v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Interpretable machine learning coupled to gene regulatory networks uncovers subcircuits underlying cell fate decisions
AU - Z. H. Rarani
AU - S. Keshari
AU - A. Sachan
AU - A. Saini
AU - N. A. Pease
AU - J. Fan
AU - A. Fu
AU - Y. C. Liu
AU - A. U. Gurkar
AU - G. M. Delgoffe
AU - H. Singh
AU - J. Das
PY - 2026
DA - 2026/09/28
PB - bioRxiv
DO - 10.64898/2026.09.28.754797
UR - https://www.biorxiv.org/content/10.64898/2026.09.28.754797v1
AB - Gene regulatory networks (GRNs) model causal linkages that control cell fate decisions and differentiation transitions. Prioritizing regulatory subnetworks underlying cell state differences is of critical importance, but current methods including those reliant on topological metrics introduce circularity as the metrics prioritizing TFs are computed from the same networks whose assumptions they inherit. Separately, interpretable machine learning methods can identify latent factors (LFs) that discriminate cellular states with formal statistical guarantees but do not model regulatory linkages. Here, we present FOCAL (Factor-Outcome Coupling for Assessment of Linkages), a paradigm to prioritize regulatory subnetworks by coupling state-specific and dynamic GRNs with outcome-supervised LFs learned using interpretable machine learning without reference to network topology. This shifts GRN focus from macroscopic TF nodes to state-specific and dynamic TF-gene linkages. In B and T cells, FOCAL identified GIFs (GRNs coupled to Interpretable latent Factors), prioritized regulatory subnetworks underlying established states as well as transient regulatory episodes preceding them. By coupling LFs learnt from perturbation experiments of lineage-defining TFs, FOCAL identified transcriptional predisposition to alternative fates within progenitor cell populations before overt differentiation. This uncovered a novel NFATC2-IRF8 interplay in activated B cells, that was validated by in-vitro and in-vivo genetic perturbations. The two transcription factors act cooperatively to restrain extrafollicular plasmablast differentiation and promote germinal center B cell fate.
ER -