Clustering gene expression time series data using an infinite Gaussian process mixture model | JoVE Visualize

Area of Science:

Genomics
Computational Biology
Systems Biology

Background:

Transcriptome-wide time series expression profiling is crucial for understanding cellular responses to environmental stimuli.
Clustering genes with similar temporal expression patterns is a key initial step in analyzing such data.
Existing methods may not fully capture complex temporal dependencies and cluster structures.

Purpose of the Study:

To introduce a novel nonparametric model-based method, Dirichlet process Gaussian process mixture model (DPGP), for analyzing time series gene expression data.
To jointly model data clusters and temporal dependencies for improved accuracy in identifying gene expression patterns.
To provide a robust computational tool for uncovering shared regulatory mechanisms in biological responses.

Main Methods:

Developed a Dirichlet process Gaussian process mixture model (DPGP) that integrates Dirichlet processes for cluster modeling and Gaussian processes for temporal dependencies.
Validated DPGP's accuracy using hundreds of simulated datasets and comparing it against state-of-the-art clustering approaches.
Applied DPGP to analyze published microbial stress response microarray data and novel human cell line RNA-seq data exposed to dexamethasone.

Main Results:

DPGP demonstrated superior accuracy in clustering gene expression data compared to existing methods on simulated datasets.
The method successfully identified biologically relevant gene clusters in both microbial and human experimental datasets.
Validation using transcription factor binding and histone modification data supported the biological relevance of the identified clusters.

Conclusions:

Jointly modeling cluster number and temporal dependencies using DPGP enhances the characterization of cellular responses.
The DPGP method effectively reveals shared regulatory mechanisms underlying gene expression dynamics.
DPGP offers a powerful and accurate tool for the analysis of time series transcriptomic data, with software freely available.