Switched Latent Force Models for Reverse-Engineering Transcriptional Regulation in Gene Expression Data

Area of Science:

• Systems Biology
• Computational Biology
• Molecular Biology

Background:

• Gene expression regulation is crucial for cellular survival and response to environmental changes.
• Transcription factors (TFs) mediate this regulation, but their dynamics are challenging to measure experimentally.
• Existing differential equation models struggle with the rapid, discontinuous changes in TF activity.

Purpose of the Study:

• To develop a novel computational framework for reverse-engineering transcriptional regulation.
• To enable exact inference of latent TF activities driving gene expression.
• To accurately model sudden changes and non-linearities in gene regulatory networks.

Main Methods:

• A switched dynamical latent force model was developed.
• The model uses linear differential equations to describe TF activities.
• A switching approach was introduced to handle discontinuities in TF dynamics.

Main Results:

• The model successfully infers continuous-time TF profiles from gene expression data.
• It accurately fits both simulated and real biological data.
• The framework demonstrates versatility in capturing discrete changes and non-linearities.

Conclusions:

• The proposed model offers a powerful tool for understanding gene expression dynamics.
• It overcomes limitations of traditional modeling approaches for TF regulation.
• This framework advances the study of complex biological regulatory networks.