How a few hundred pieces
become 10⁸ receptors
One T cell recognises exactly one thing. Yet your body must hold receptors for pathogens it has never met.
These 20 lessons work that design problem through in three stages — making, selecting, reading. And not just in words: you open four real human repertoires, 11,791 cells, published by 10x Genomics, and compute V gene usage, CDR3 lengths, diversity indices, clonal expansion, public clones and antigen specificity yourself.
Including working out, from the data alone, that two of the samples are the same person. Python runs right inside this page.
How you'll learn
Not a simulation. The actual receptor sequences of four real people, 11,791 cells, published by 10x Genomics. Even the IMGT germline gene counts are counted from the file.
Two of the four samples are the same person. We don't tell you which — you work it out in lesson 11, and see why that method works while another one doesn't.
You cannot compare diversity across samples of different size. Look at fewer cells and clonal expansion disappears. Failed rearrangements are simply absent from the data. You train, repeatedly, to think about what isn't there.
Lesson 18 lists eight statements this data supports and six it does not, each with its reason — and what you would need to add to turn each ✕ into a ✓.