A multi-Poisson dynamic mixture model to cluster developmental patterns of gene expression by RNA-seq

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Dynamic gene expression is crucial for organism response to development and environment.
• RNA sequencing (RNA-seq) generates extensive temporal gene expression data, posing challenges for functional classification.
• Understanding gene function requires effective analysis of complex expression patterns.

Purpose of the Study:

• To develop a clustering mixture model for discovering gene groups based on RNA-seq data during organ development.
• To accommodate the discrete nature of RNA-seq read counts and model temporal gene expression dynamics.
• To provide a computational tool for analyzing gene expression patterns and understanding regulatory mechanisms.

Main Methods:

• Developed a clustering mixture model integrating multivariate Poisson distribution for RNA-seq count data.
• Incorporated a first-order autoregressive process to model temporal dependence in gene expression.
• Utilized the Expectation-Maximization algorithm for parameter estimation and model selection.

Main Results:

• The model successfully identified distinct groups of genes with specific expression patterns during organ development.
• Demonstrated the model's application on real RNA-seq data from white poplar catkin development.
• Validated the model's usefulness and accuracy through computer simulations.

Conclusions:

• The developed clustering mixture model is a valuable tool for analyzing RNA-seq data.
• Facilitates a global view of gene expression dynamics and enhances understanding of gene regulation.
• Offers a robust method for classifying genes based on their temporal expression profiles.