iscGPS

iscGPS infers transitioning potential between clusters in single-cell transcriptomic data to identify cell states and their transcriptional and lineage relationships.

Key Features:

• Clustering with SCORE: Decomposes mixed cell populations from one or more samples into distinct clusters using the SCORE (Single Cell Clustering Optimization and Evaluation) algorithm.
• Transition estimation with scGPS algorithm: Estimates transitioning potential between pairs of clusters using a machine learning classification approach implemented as the scGPS algorithm.
• Feature selection via network and modeling approaches: Identifies biological processes and candidate driver genes that connect different cell populations through network- and model-based feature selection.
• Implementation: Computational implementation in R with performance optimization using C++.

Scientific Applications:

• Developmental biology: Infers lineage relationships and cell state transitions during developmental processes from single-cell transcriptomic data.
• Cell differentiation and lineage tracing: Maps trajectories and potential differentiation paths between identified cell clusters.
• Cellular plasticity and driver-gene discovery: Detects cellular plasticity dynamics and candidate driver genes underlying transitions between cell populations.

Methodology:

Decomposition of mixed populations with SCORE; pairwise transition potential estimation via a machine learning classification approach implemented as scGPS; feature selection using network- and model-based methods; implemented in R with C++ optimization.