CellTrails

CellTrails uncovers, analyzes, and visualizes single-cell gene-expression dynamics by reconstructing de novo chronological ordering and branching trajectories from single-cell expression data.

Key Features:

• Trajectory Reconstruction: CellTrails reconstructs branching trajectories spanning from progenitor cells to distinct phenotypes using de novo chronological ordering of single-cell expression data.
• Manifold Learning: CellTrails employs a geometrically motivated concept of lower-dimensional manifold learning to counteract intrinsic noise of single-cell data caused by drop-outs, technical variance, and redundancy of predictive variables.
• Expression Dynamics Inference: CellTrails enables definition and inference of expression dynamics of individual and multiple pathways towards distinct phenotypes.
• Trail Extraction and Alignment: CellTrails extracts and aligns developmental trails and associates pseudotime with phenotypic measurements.
• Differential Trail Analysis: CellTrails performs differential trail analysis to reveal high-resolution dynamics of transcripts controlling distinct developmental outcomes.
• Visualization: CellTrails provides an intuitive graphical representation of expression patterns along all branches simultaneously.

Scientific Applications:

• Hair-Bundle Morphogenesis: CellTrails was applied to dissect sequential and overlapping cellular processes in hair-bundle growth by reconstructing a bifurcating trajectory from progenitor cells to mature striolar and extrastriolar hair cells in the developing chick utricle.
• Gene Expression-Phenotype Association: CellTrails links expression dynamics of individual genes with bundle growth stages by associating pseudotime with bundle length measurements.
• Developmental Process Dissection: CellTrails reveals high-resolution dynamics of transcripts controlling distinct phenotypic outcomes, including those encoding proteins that regulate [Ca2+]i or mediate crosslinking and lengthening of actin filaments.

Methodology:

CellTrails uses a geometrically motivated concept of lower-dimensional manifold learning for de novo chronological ordering, followed by reconstruction of branching trajectories, extraction and alignment of developmental trails, association of pseudotime with phenotypic measurements, and differential trail analysis.