SCRATCH-CellTrajectory

Introduction

SCRATCH-CellTrajectory infers differentiation trajectories and developmental pseudotime from annotated single-cell datasets. It integrates CytoTRACE2, Slingshot, and Monocle3 within a unified, containerized pipeline — enabling robust, reproducible trajectory inference across tumor and immune cell populations.

This module is a key component of the SCRATCH single-cell analysis ecosystem, fully orchestrated through Nextflow + Quarto (QMD) and supporting both Docker and Singularity execution.

Prerequisites

Nextflow ≥ 21.04.0 Java ≥ 8 Docker or Singularity/Apptainer Git

R packages (via container):

Seurat, monocle3, slingshot, tradeSeq, CytoTRACE2, ggplot2, patchwork, Matrix, dplyr, WGCNA

Installation

git clone https://github.com/WangLab-ComputationalBiology/SCRATCH-CellTrajectory.git cd SCRATCH-CellTrajectory

main.nf — primary entrypoint

modules/local/main.nf — executes QMD notebooks

subworkflows/local/SCRATCH_CellTrajectory.nf — orchestrates integration across methods

nextflow.config — defines default resources and container paths

Quick Start

Minimal example (Docker)

nextflow run main.nf -profile docker
--input_seurat_object Batch_Correction_output.RDS
--project_name CellTrajectoryDemo

Minimal example (Singularity)

nextflow run main.nf -profile singularity
--input_seurat_object Batch_Correction_output.RDS
--project_name CellTrajectoryDemo

Typical Workflow Execution

CytoTRACE2 estimates differentiation potential.

Slingshot infers lineages and pseudotime trajectories.

Monocle3 reconstructs the principal graph and refines pseudotime ordering.

Results are merged, correlated, and visualized to provide consistent developmental trajectories.

Key Parameters

--input_seurat_object Input Seurat .RDS file --project_name Project label --outdir Output directory --n_threads Number of CPU threads

Outputs

SCRATCH-CellTrajectory_output/ ├── rds/ │ ├── cytotrace_scores.rds │ ├── slingshot_lineages.rds │ ├── monocle3_cds.rds │ └── integrated_pseudotime_table.csv ├── figures/ │ ├── CytoTRACE_UMAP.png │ ├── Slingshot_UMAP.png │ ├── Monocle3_UMAP.png │ └── Correlation_CytoTRACE_vs_Monocle3.png └── report/ └── CellTrajectory.html

Deliverables:

Pseudotime trajectories from three independent algorithms Integrated comparison plots Annotated Seurat and Monocle3 objects for downstream analysis

Example Full Run

nextflow run main.nf -profile singularity
--input_seurat_object Batch_Correction_output.RDS
--project_name Lung_Trajectory
-resume

Documentation

All QMD notebooks include annotated code and figures demonstrating key steps in CytoTRACE, Slingshot, and Monocle3 pseudotime reconstruction.

License

Licensed under the GNU General Public License v3.0 (GPLv3).

Developers:

[email protected] [email protected]