PBMC scRNA-seq Analysis (Seurat, 10x PBMC3k)

This project contains an R Markdown workflow (scRNAseq.Rmd) for analysing 10x Genomics PBMC3k single-cell RNA-seq data using Seurat.
The notebook walks through data import, quality control, dimensionality reduction, clustering, and cell type annotation.

Main steps

1. Downloads

   • Install/load required R packages (Seurat, dplyr, patchwork, DT, ggplot2).
   • Download and unpack the PBMC3k count matrix from 10x Genomics.

2. Preprocessing & QC

   • Create a Seurat object.
   • Compute standard QC metrics (nFeature_RNA, nCount_RNA, percent.mt).
   • Visualise QC metrics and filter out low-quality cells (feature/UMI cutoffs, high mitochondrial content).

3. Normalisation & Feature Selection

   • Log-normalise counts (NormalizeData).
   • Identify highly variable genes (FindVariableFeatures).

4. Dimensionality Reduction & Clustering

   • Scale data and run PCA.
   • Inspect PCs via loadings, heatmaps, and elbow plot.
   • Build a neighbour graph and compute UMAP embeddings.
   • Explore the effect of number of PCs and clustering resolution, then fix on 10 PCs, resolution 0.3.

5. Marker Detection & Cell Type Annotation

   • Use FindAllMarkers (ROC test, only positive markers) to identify cluster markers.
   • Build a composite marker score (AUC × log₂ fold change × specificity) and select one top marker per cluster.
   • Visualise marker expression on UMAP and assign labels:
     • Naive CD4 T, CD14⁺ Mono, NK/CD8 T, B cells, CD16⁺ Monocytes, Dendritic cells, Platelets.

Files in this repository

• scRNAseq.Rmd - R Markdown notebook containing the full PBMC scRNA-seq analysis workflow (QC, normalization, PCA, UMAP, clustering, marker detection, and cell type annotation).
• scRNAseq.html - Rendered HTML report generated from scRNAseq.Rmd, with all figures, tables and narrative results.