redPATH

redPATH reconstructs pseudo-developmental time from single-cell RNA sequencing (scRNA-seq) data by formulating cell ordering as a Hamiltonian path problem to infer progression through cellular states.

Key Features:

• Consensus asymmetric Hamiltonian path algorithm: Formulates cell ordering as a Hamiltonian path problem and applies a consensus asymmetric Hamiltonian path strategy to infer pseudo-temporal orderings.
• Directional modeling: Models asymmetric relationships between cells to capture the directional nature of developmental processes.
• Trajectory visualization: Provides visualization components for interpreting inferred lineage trajectories and differentiation dynamics.
• Cross-dataset validation: Evaluated on multiple datasets, including neural stem cell and cancer scRNA-seq data, and reported to segment distinct developmental stages.

Scientific Applications:

• Glioma stem-like subpopulation identification: Applied to malignant glioma scRNA-seq data to identify a stem cell-like subpopulation associated with GFAP, ATP1A2, IGFBPL1, and ALDOC expression and reduced expression of the quiescent marker ID3.
• Tumor gene prioritization: Highlighted genes implicated in tumor biology, including MCL1 (apoptosis regulation) and CSF1R (macrophage reprogramming linked to tumor growth modulation).
• Development and disease progression analysis: Reconstructs pseudo-developmental timelines to characterize differentiation programs, identify subpopulations, and prioritize candidate regulatory genes in developmental biology and cancer research.

Methodology:

Formulates cell ordering as a Hamiltonian path problem and infers pseudo-temporal orderings using a consensus asymmetric Hamiltonian path algorithm that models asymmetric cell–cell relationships.