GPseudoClust

GPseudoClust jointly infers pseudotemporal ordering and gene clusters from single-cell RNA sequencing (scRNA-seq) mRNA expression data while quantifying uncertainty in both pseudotime and clustering.

Key Features:

• Joint Inference: Performs simultaneous estimation of pseudotime and gene clustering rather than treating them as separate steps.
• Uncertainty Quantification: Quantifies uncertainty in both pseudotemporal ordering and gene cluster assignments.
• Methodological Integration: Combines a pseudotime inference method with non-parametric Bayesian clustering.
• Efficient Computation: Employs Markov Chain Monte Carlo (MCMC) sampling and subsampling strategies to improve computational efficiency.
• Versatility Across Datasets: Has been demonstrated on simulated and experimental datasets for analysis of temporal gene expression.

Scientific Applications:

• Developmental Biology: Identifies gene clusters by temporal expression patterns to study regulatory mechanisms in cell differentiation and development.
• Cancer Research: Detects gene clusters associated with tumor heterogeneity and progression across stages of cancer development.
• Stem Cell Research: Maps differentiation trajectories and lineage specification by clustering temporally varying genes.

Methodology:

Combines pseudotime inference with non-parametric Bayesian clustering, uses Markov Chain Monte Carlo (MCMC) sampling and subsampling strategies, estimates temporal ordering of cells from scRNA-seq mRNA expression, and groups genes without assuming a predefined number of clusters.