Clumped-MCEM: Inference for multistep transcriptional processes.

Area of Science:

• Biochemistry
• Computational Biology
• Systems Biology

Background:

• Multistep biochemical reactions, such as transcriptional processes, are fundamental to biological systems.
• Simplifying complex reaction pathways is crucial for accurate modeling and analysis.
• Existing methods for analyzing gene expression data can be computationally intensive.

Purpose of the Study:

• To develop a novel, efficient computational method for simulating and inferring kinetic parameters in multistep biochemical reactions.
• To introduce a model reduction strategy that simplifies complex reaction states while incorporating time delays.
• To validate the new method, Clumped-MCEM, using time-series data from the mouse glutaminase promoter.

Main Methods:

• Devised a model reduction strategy representing multiple OFF states as a single state with a time delay for burst frequency.
• Developed Clumped-MCEM (Clumped Markov Chain Monte Carlo Expectation-Maximization) for simulation and parameter inference.
• Applied Clumped-MCEM to time-series data of the endogenous mouse glutaminase promoter for validation and parameter estimation.

Main Results:

• Clumped-MCEM demonstrated higher efficiency for time-series data analysis compared to existing methods.
• The method achieved comparable numerical accuracy to state-of-the-art approaches (Bursty MCEM^2 and delay Bursty MCEM).
• Clumped-MCEM significantly reduced computational cost, by 57.58% versus Bursty MCEM^2 and 32.19% versus delay Bursty MCEM.

Conclusions:

• Clumped-MCEM provides an efficient and accurate approach for kinetic parameter inference in complex biological systems.
• The model reduction strategy effectively simplifies multistep reactions, improving computational performance.
• This method offers a valuable tool for analyzing gene expression dynamics and understanding transcriptional regulation.