CALISTA

CALISTA performs probabilistic analysis of single-cell transcriptomic data to cluster cells, reconstruct developmental lineages, identify transition genes, and order cells in pseudotime for the study of cell differentiation.

Key Features:

• Single-Cell Clustering: Groups individual cells by their gene expression profiles to identify distinct cellular populations within heterogeneous samples.
• Reconstruction of Cell Lineage Specification: Reconstructs cell lineage trees to reveal developmental pathways and differentiation relationships among cells.
• Transition Gene Identification: Identifies genes that change expression during cellular transitions and contribute to cell fate decisions.
• Cell Pseudotime Ordering: Orders cells along pseudotemporal trajectories to represent progression of cellular states over time.
• Likelihood-Based Probabilistic Modeling: Models single-cell mRNA counts using a likelihood-based probabilistic distribution that accounts for stochastic gene transcriptional bursts and random technical dropout events.
• Numerical Efficiency and Scalability: Provides computational performance and scalability suitable for datasets ranging from a few hundred to tens of thousands of cells.
• End-to-End Pipeline for Differentiation Studies: Implements an analytical pipeline tailored for analyses focused on cell differentiation.
• Cross-Technology Applicability: Demonstrated on diverse single-cell gene expression datasets derived from various single-cell transcriptional profiling technologies.

Scientific Applications:

• Cell Differentiation Analysis: Dissects transcriptional programs and cell-state transitions during differentiation.
• Developmental Pathway Reconstruction: Maps developmental lineages and branching events in heterogeneous cell populations.
• Transition-Gene Discovery: Detects candidate regulatory genes implicated in lineage transitions and fate decisions.
• Pseudotemporal Dynamics: Analyzes dynamic processes such as progression of cellular states over pseudotime.
• Cross-Platform Single-Cell Studies: Applies to gene expression datasets generated by multiple single-cell transcriptional profiling technologies.

Methodology:

CALISTA uses a likelihood-based approach that models single-cell mRNA counts with a probabilistic distribution explicitly accounting for stochastic gene transcriptional bursts and random technical dropout events.