CellChat

Stages

1. 1.
   Single-cell preprocessing and clustering(broad_screen)

   This stage organizes the single-cell dataset into analyzable cell populations and subsets for downstream interpretation.

   Selection: Quality control, integration, normalization, clustering, and marker analysis of scRNA-seq data using Seurat.

2. 2.
   Epithelial CNV and malignant-state inference(functional_characterization)

   This stage characterizes epithelial malignant potential using inferred CNV patterns.

   Selection: inferCNV is applied to infer CNV heterogeneity and malignant status among epithelial cells.

3. 3.
   Pseudotime ordering of epithelial progression(secondary_characterization)

   This stage places epithelial states into an inferred progression trajectory during malignant development.

   Selection: Monocle2 is used for pseudotemporal ordering of epithelial deterioration and lineage progression.

4. 4.
   Cell-cell communication analysis(secondary_characterization)

   This stage identifies putative signaling interactions among cell compartments in the tumor microenvironment.

   Selection: CellChat is used to analyze communication across epithelial, fibroblast, and other tumor microenvironment compartments.

5. 5.
   Pathway activity analysis(secondary_characterization)

   This stage summarizes transcriptomic differences as pathway activity programs across identified groups.

   Selection: GSVA is used for pathway activity analysis across epithelial subsets and CNV-defined groups.

Stages

1. 1.
   Interaction database construction(library_design)

   This stage provides the interaction knowledge base needed to accurately represent cell-cell signaling links before inference from scRNA-seq data.

   Selection: interactions among ligands, receptors, and cofactors that represent known heteromeric molecular complexes

2. 2.
   Communication network inference and analysis from scRNA-seq(functional_characterization)

   This stage uses the interaction database and scRNA-seq data to infer communication networks and analyze signaling roles of cells.

   Selection: quantitative inference and analysis of intercellular communication networks from scRNA-seq data

3. 3.
   Cross-dataset pathway classification and contrast(secondary_characterization)

   This stage classifies signaling pathways and distinguishes conserved from context-specific pathways across datasets.

   Selection: manifold learning and quantitative contrasts across datasets

4. 4.
   Application to skin datasets(confirmatory_validation)

   This stage demonstrates the tool on biological datasets to show that it can extract complex signaling patterns.

   Selection: ability to extract complex signaling patterns in mouse and human skin datasets

Steps

1. 1.
   Construct interaction database of ligands, receptors, and cofactorsinteraction knowledge base

   Provide an accurate representation of known signaling links including heteromeric molecular complexes.

   The abstract presents database construction before tool development, implying it is the prerequisite knowledge layer for downstream inference.

2. 2.
   Develop CellChat to infer and analyze intercellular communication networks from scRNA-seq datacomputational inference tool

   Quantitatively infer and analyze intercellular communication networks.

   This follows database construction because the inference tool depends on accurate interaction representation.

3. 3.
   Predict signaling inputs and outputs and analyze coordination using network analysis and pattern recognitionanalysis engine

   Identify major signaling inputs and outputs for cells and how cells and signals coordinate for functions.

   This is a downstream analysis on inferred communication structure to extract interpretable functional organization.

4. 4.
   Classify signaling pathways and delineate conserved and context-specific pathways across datasetscross-dataset analysis tool

   Compare signaling pathways across datasets to identify conserved and context-specific patterns.

   This occurs after communication inference because pathway classification and contrasts require inferred signaling structure from one or more datasets.

5. 5.
   Apply CellChat to mouse and human skin datasetsapplied analysis tool

   Demonstrate the ability to extract complex signaling patterns in biological datasets.

   This confirmatory application follows method development and analysis design to show practical utility on real datasets.