CellFateGenie Analysis

Use this skill when the user wants to identify genes that drive cell fate decisions along a developmental trajectory. CellFateGenie discovers pseudotime-associated genes using adaptive ridge regression and then scores lineage-specific fate-driving genes via manifold density estimation.

This skill is used after trajectory inference — the user must already have pseudotime values computed (e.g., from Palantir, VIA, or diffusion pseudotime).

Overview

CellFateGenie answers: "Which genes change most significantly along pseudotime, and which are specifically driving a particular lineage?" It works in two phases:

1. Gene selection — Adaptive Threshold Regression (ATR) iteratively removes low-coefficient genes while monitoring R² to find the minimal gene set that explains pseudotime
2. Lineage scoring — Mellon density estimation on the manifold identifies low-density transition regions, and lineage-specific variability scoring pinpoints fate-driving genes

Prerequisites

• Pseudotime: Must exist as a column in adata.obs. Compute first using Palantir, VIA, DPT, or any trajectory method.
• Mellon (optional but important): pip install mellon for density estimation. Without it, low_density() will fail.
• Expression data: Works on the .X matrix. Log-normalized data is fine (unlike SCENIC which needs raw counts).