CSHMM

CSHMM models time-series single-cell RNA-Seq (scRNA-Seq) data to reconstruct continuous developmental trajectories and probabilistic branching topology for studying cellular differentiation and gene expression dynamics.

Key Features:

• Probabilistic Branching Model: Employs a probabilistic branching framework that represents branching structure in developmental processes using continuous-state Hidden Markov Models (HMMs).
• Continuous-State Representation: Represents cell states on a continuous scale to assign cells along developmental trajectories with higher resolution than discrete models.
• Efficient Learning and Inference Algorithms: Includes algorithms for learning model parameters and performing inference to assign cells to trajectories from time-series scRNA-Seq data.
• Improved Accuracy in Trajectory Reconstruction: Uses continuous-state modeling to improve reconstruction of branching topology and continuous assignment of cells along developmental paths.

Scientific Applications:

• Developmental Biology: Reconstructs lineage relationships and identifies transition points in cellular differentiation from time-series scRNA-Seq.
• Gene Expression Analysis: Facilitates identification of known and novel gene markers by analyzing gene expression patterns across continuous cell assignments.

Methodology:

Defines a continuous-state Hidden Markov Model (HMM) tailored to time-series scRNA-Seq data and develops efficient algorithms for learning model parameters and performing inference to reconstruct trajectories and assign cells.