Title: Machine learning-assisted decoding of temporal transcriptional dynamics via fluorescent timer

 

Abstract:

Investigating the temporal dynamics of gene expression is crucial for understanding gene regulation across various biological processes. Using the Fluorescent Timer protein, the Timer-of-cell-kinetics-and-activity (Tocky) system enables analysis of transcriptional dynamics at the single-cell level. However, the complexity of Timer fluorescence data has limited its broader application.

 

In this talk, I will present an integrative approach that combines molecular biology and machine learning to analyse Foxp3 transcriptional dynamics using flow cytometric Timer data. We developed a convolutional neural network (CNN)-based method incorporating image transformation and class-specific feature visualization to identify regulatory patterns at the single-cell level. Using a CRISPR-engineered mutant Foxp3-Tocky mouse model lacking the enhancer Conserved Non-coding Sequence 2 (CNS2), we reveal enhancer-specific control of transcription frequency under immune stimulation.

 

In addition, our analysis of wild-type Foxp3-Tocky mice across developmental stages demonstrates age-dependent shifts in Foxp3 expression dynamics, particularly highlighting thymus-like transcriptional profiles in neonatal peripheral T cells.

 

In conclusion, this work establishes a framework integrating CRISPR mutagenesis, single-cell time-resolved analysis, and machine learning to decode transcriptional regulation in vivo.

 

Reference: Irie, N., Takeda, N., Satou, Y., Araki, K. & Ono, M. Machine learning-assisted decoding of temporal transcriptional dynamics via fluorescent timer. Nature Communications 16, 5720 (2025). https://doi.org/10.1038/s41467-025-05720-6