🧬 Genomics & DNA

What you’ll learn

1. A Text, Not a BlueprintReplace the blueprint metaphor with the right one — a text — and separate genomics (what do the letters say?) from genetics (what gets inherited?).Your genome is not a scale drawing with parts that correspond to parts of you; it is a three-billion-letter string in a four-letter alphabet. The completely sequenced human genome published in 2022 came to 3,054,815,472 base pairs of nuclear DNA. Genomics asks what the text says rather than what is inherited — and the whole course lives in the widening gap between being able to read it and being able to understand it.
2. Sanger's Trick: Stopping the Copy on PurposeUnderstand chain-termination sequencing as an act of deliberate sabotage, and see why the answer is inferred from fragment lengths rather than observed.You cannot look at a DNA molecule, so Sanger's 1977 method doesn't try: it copies the strand with a small proportion of sabotaged letters that lack the hook the next letter needs, freezing each copy at a random position. Sorting the resulting fragments by length in an electric field produces a ladder whose rungs differ by exactly one letter. Reading which of four lanes each rung sits in gives the sequence — the order inferred entirely from the lengths of the broken pieces.
3. You Can Only Read It in PiecesSee why shotgun sequencing works, why read depth is an error check rather than mere coverage, and why repetitive sequence defeats it.No method reads three billion letters in one pass, so the genome is shredded across many copies, read as millions of fragments, and reassembled from overlaps. Randomness is the mechanism, not a compromise: overlapping breaks are the only evidence of adjacency, and reading each letter roughly thirty times over is how a real variant is distinguished from a machine's typo. The approach collapses in repeats, where every fragment overlaps every other equally well — which is why about 8% of the human genome stayed unread for nineteen years.
4. The RaceUnderstand the public–private race as a genuine technical disagreement about repeats, and see why the public project's data policy mattered more than who won.The Human Genome Project launched in 1990 as a public consortium that released its data freely and continuously; in 1998 Celera Genomics proposed shotgunning the whole genome without a map. The objection to Celera was technical — without anchoring, repeats would produce an assembly that looked complete and was subtly scrambled — and the accounts published by the principals still differ sharply on who owed what to whom. The 2000 announcement was a truce; the lasting result was that competition compressed the timetable and the reference text ended up free to everyone forever.
5. The Price of a LetterTrace the collapse in sequencing cost using NHGRI's own figures, and understand why a price change altered which biological questions were askable.NHGRI has tracked the production cost of sequencing a human genome since 2001: about $95.3 million in September 2001, about $7.1 million by October 2007, then about $342,500 twelve months later once short-read machines arrived, and about $525 by May 2022 — a fall of roughly 180,000-fold in twenty-one years. The point is not that things got cheaper but that the questions changed: at $95 million a genome is a national project, at $525 it is a lab consumable and you can study variation across a hundred thousand people. Our ability to produce sequence outran our ability to interpret it, and never caught up.
6. The Sweepstake Nobody WonUse the gene-count surprise to dismantle the assumption that biological complexity is stored in the size of the parts list.GeneSweep, a betting pool run during the sequencing, drew more than 460 guesses at the human protein-coding gene count; the winning bet of 25,947 was the lowest of them all and still too high. GENCODE's current reference annotation lists 19,442 protein-coding genes — roughly a roundworm's count — alongside 644,292 transcripts, 35,885 long non-coding RNA genes and 14,702 pseudogenes. Complexity lives not in the number of parts but in how many combinations they are used in and in the vast apparatus deciding when, where and how much.
7. The Fight About JunkSee that the ENCODE controversy was a dispute over the definition of 'functional', and that measuring activity exhaustively cannot answer a question about meaning.ENCODE's 2012 survey reported that around 80% of the human genome is functional, using a biochemical definition — transcribed, protein-bound, or chemically marked. Critics led by Dan Graur argued that activity is cheap and noisy and that function should mean maintained by selection, a standard under which less than 10% qualifies. Neither headline survived: 'it's all junk' was wrong because the regulatory apparatus is far larger than anyone expected, and '80% is functional' was a claim about activity dressed as a claim about purpose. You cannot resolve a question about meaning by collecting more text.
8. Reading Is Not UnderstandingMake the reading/understanding gap concrete through GWAS and the Variant of Uncertain Significance.A perfectly sequenced genome tells you a great deal about a few strong single-gene variants and very little about height, heart disease or depression, where influence is spread across thousands of variants entangled with environment. GWAS reliably finds statistical associations without explaining them — a hit is a fingerprint with no suspect, often nowhere near a gene. In the clinic the same gap is called a Variant of Uncertain Significance, and its rate is markedly higher for people whose ancestry is under-represented in reference databases.
9. Where Genomics Actually DeliversEstablish the rule that predicts every success and every disappointment: how short is the causal chain between a letter and the outcome?Genomics is transformative wherever one variant clearly breaks one protein and plainly causes one condition — trio sequencing that ends a family's years-long diagnostic odyssey, tumour-versus-healthy-tissue comparison that names a driver mutation and sometimes a drug for it, and pharmacogenomic tests that read one letter to avoid one harm. It is nearly mute wherever a trait is the sum of ten thousand tiny nudges plus a lifetime of environment. The useful question is never whether genomics is powerful but how short the chain is.
10. Ancestry, and the Story You Get ToldDistinguish what consumer genetic tests actually measure from what they report, and confront the fact that genetic privacy is not individually held.Most consumer ancestry tests genotype several hundred thousand pre-chosen positions rather than sequencing a genome; the relative-matching this enables is real and solid, while the percentage breakdown is computed by comparison against reference panels and therefore moves when the panels do. The Golden State Killer case in April 2018 showed that a genome is not individually private: distant relatives who uploaded their own results, and who had never met the suspect, were what exposed him. Reasonable people disagree sharply about what should follow, and both sides argue from real principles.
11. Finishing the JobClose the loop: long reads finally read the repeats, and finishing the reference exposed why a single reference was the wrong idea all along.The 8% missing from the 2003 reference was not scattered — it was the centromeres, satellite arrays and short chromosome arms where short-read assembly hits a wall of identical sequence. Reads tens of thousands of letters long span an entire repeat array, and on 31 March 2022 the Telomere-to-Telomere consortium published 3,054,815,472 base pairs gap to gap, adding or correcting 238 million bases, 182 million of them entirely new, and annotating 19,969 protein-coding genes. Finishing it made the deeper flaw undeniable: a single reference text turns under-representation into clinical ambiguity, which is what the pangenome exists to fix.

Questions this course answers

Grounded in trusted sources

• Sergey Nurk et al. — "The complete sequence of a human genome", Science 376:6588 (31 March 2022)
• NHGRI — "DNA Sequencing Costs: Data" (Wetterstrand, K.A.), cost-per-genome table, data through May 2022
• GENCODE Release 50 — human gene annotation statistics (gencodegenes.org)
• International Human Genome Sequencing Consortium — "Initial sequencing and analysis of the human genome", Nature 409 (2001)
• J. Craig Venter et al. — "The sequence of the human genome", Science 291 (2001)
• ENCODE Project Consortium — "An integrated encyclopedia of DNA elements in the human genome", Nature 489 (2012)
• Dan Graur et al. — "On the immortality of television sets: 'function' in the human genome according to the evolution-free gospel of ENCODE", Genome Biology and Evolution 5:3 (2013)
• F. Sanger, S. Nicklen & A.R. Coulson — "DNA sequencing with chain-terminating inhibitors", PNAS 74:12 (1977)

Every Wunder lesson is built from real, reputable sources — never invented.

Related Science courses