1. Dissociation: from tissue to a cell suspension
How enzymes and force turn tissue into single cells, and the marks this leaves in the data.
Single-cell RNA-seq
From a piece of tissue to a cell atlas. Data: 10x Genomics 3k PBMCs (2,700 cells)
Bench & sequencing
2. Droplets: one cell, one reaction chamber
How 10x Chromium packs cells and gel beads into droplets, and where empty droplets and doublets come from.
3. Cell barcodes, UMIs and reverse transcription
How each mRNA gets tagged with which cell it came from and which molecule it is.
4. Library prep and sequencing: anatomy of a read
cDNA becomes a sequencing library; what Read 1 and Read 2 each carry.
From reads to data
5. Cell Ranger: from reads to a count matrix
Alignment, barcode correction, UMI counting and cell calling, and what each web-summary metric means.
Analysis
6. Quality control: which cells to keep
Spot low-quality cells by UMIs, genes and mitochondrial fraction; drag the thresholds on real data.
7. Normalization: removing sequencing depth
Why raw counts cannot be compared directly, and what LogNormalize and SCTransform do.
8. Variable genes and PCA: capturing the main variation
Pick the most informative genes, then compress thousands of dimensions into a few dozen components.
9. Clustering: finding communities in a neighbour graph
The kNN graph and Leiden, and how resolution sets the number of clusters.
10. UMAP: a map, not a ruler
What a 2-D embedding keeps and what it loses: how to read it, and how not to.
11. Marker genes and cell-type annotation
From differential genes to cell identity: annotate pbmc3k yourself.
12. Beyond: batches, trajectories, communication, outcome
The ideas behind integration, trajectories, velocity, communication, CNV and survival, and where to continue in OmicOne.