Identification of regulatory modules by co-clustering latent variable models: stem cell differentiation

Area of Science:

• Stem cell biology
• Genomics
• Computational biology

Background:

• Directing stem cell differentiation to specific cell types is a critical challenge.
• Previous studies identifying gene regulators lacked a systemic view of regulatory modules.
• A unified description of regulatory modules is needed for understanding differentiation.

Purpose of the Study:

• To characterize major stem cell species and elucidate cell type-specific transcription factors.
• To develop a systemic view of regulatory modules governing stem cell differentiation.
• To apply a novel analytical framework to stem cell differentiation problems.

Main Methods:

• Employed a co-clustering latent variable model (LVM).
• Utilized gene expression profiles and upstream genomic sequences from 21 stem cell subpopulations.
• Identified regulatory modules and their targeted genes within stem cell clusters.

Main Results:

• Uncovered cell type-specific transcription factors and regulatory modules, such as GABP and E2F for proliferation, and Ap2alpha and Ap2gamma for quiescence.
• Demonstrated that constituent genes targeted by modules effectively revealed stem cell cluster identities.
• Validated the analytical framework through a detailed case study of stem cell differentiation.

Conclusions:

• The LVM-based method provides a powerful approach to elucidate regulatory modules and cell type-specific transcription factors in stem cell differentiation.
• The analytical framework is applicable to diverse problems with similar characteristics in stem cell biology.
• This study offers a unified description of regulatory modules, advancing the understanding of stem cell differentiation mechanisms.