How mohamednhassan/sat_atlas works
Project Type and Purpose
• This repository is an open-source single-cell transcriptomics analysis pipeline targeting the vascular niche of human adipose tissue.
• It combines R and Python scripts (RMarkdown & Jupyter notebooks) to reproduce data processing, integration, differential expression, functional annotation, compositional analyses, cell–cell communication inference, cross-species mapping, and pseudotime trajectory inference.
File Structure
Top-level directories map to major pipeline stages:
• Data_processing – RMarkdown pipelines for QC, normalization, integration, cell‐type annotation.
• Analysis – DE analysis, gene ontology, scoring, regulatory network inference (pySCENIC, scDRS).
• CompositionalAnalysis – scCODA compositional shifts.
• Connectome – cell–cell communication (NicheNet automation, LIANA+).
• Cross-species – cell mapping and variational integration (scMAP, scVI for mouse & pig).
• Pseudotime_trajectory – multiple trajectory tools (CellRank, PAGA, Palantir).
README Insights
• Confirms use of both R and Python.
• Focuses on reproducibility via literate programming (Rmd & notebooks).
• No dedicated frontend/backend—entire repository is a modular computational pipeline.
Main Components and Their Relationships
a. Raw Input Data
– Single‐cell count matrices (human SAT samples) stored externally or in data folder.
b. Compute Environment
– R (Seurat, scCODA, CellRank, etc.), Python (scanpy, scVI, pySCENIC, etc.), Jupyter, RStudio.
c. Data Processing Module
– Sequential RMarkdown scripts → produce processed Seurat/AnnData objects.
d. Analysis Branches (all consume processed objects)
– Differential Expression & Functional Annotation (Analysis/)
– Compositional shifts (CompositionalAnalysis/)
– Connectome inference (Connectome/)
– Cross‐species mapping (Cross-species/)
– Pseudotime trajectories (Pseudotime_trajectory/)
e. Outputs
– Figures, tables, processed RDS/AnnData files, interactive notebooks.
f. Version Control & Collaboration
– GitHub repository, LICENSE.
Architectural Patterns & Design Principles
• Modular pipeline: each stage is self-contained under its directory.
• Data artifact passing: standardized Seurat/AnnData objects flow between stages.
• Polyglot environment: explicit separation of R vs. Python workflows.
• Reproducibility via literate programming: RMarkdown & Jupyter.