Bayesian inference based modelling for gene transcriptional dynamics by integrating multiple source of knowledge.

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• Identifying genome-wide transcriptional regulatory networks is crucial in the post-genomic era.
• Existing methods often rely on qualitative models, lacking quantitative insights into regulatory systems.
• There is a need for models that can quantitatively describe the interactions between transcription factors and target genes.

Purpose of the Study:

• To propose a quantitative model for transcriptional regulatory networks based on binding affinity.
• To incorporate multiple factors like binding affinity and transcription factor activity into a learning model.
• To bridge the gap between nucleotide frequency and gene transcription rates.

Main Methods:

• Developed a binding affinity-based regulatory model.
• Integrated transcription factor (TF) binding affinity and activity levels into a general learning framework.
• Utilized promoter sequence features and nucleosome occupancy to estimate regulator binding probabilities.

Main Results:

• The proposed model quantifies transcriptional regulatory networks.
• It effectively estimates TF activity and kinetic parameters.
• The model bridges the gap between background nucleotide frequency and gene transcription rates, outperforming previous models using only microarray data.

Conclusions:

• The binding affinity-based model provides a quantitative approach to transcriptional regulatory networks.
• Experimental validation on microarray datasets demonstrates effective parameter and TF activity identification.
• The introduced kinetic parameters offer greater biological interpretability compared to existing models.