Reading,
Writing,
Editing Life.

In April 1953 a single page in Nature proposed the double helix. Watson and Crick — drawing on Rosalind Franklin's Photo 51 and Maurice Wilkins's diffraction work at King's College — rewrote heredity as a chemical structure: two anti-parallel sugar-phosphate strands, hydrogen-bonded base pairs, A-T and G-C, twisting at roughly 10.5 base pairs per turn.

Within a decade Marshall Nirenberg and others had cracked the genetic code; in 1965 Robert Holley solved the structure of yeast alanine tRNA. The metaphor was now infrastructure.

Stanford's Stanley Cohen and UCSF's Herbert Boyer met at a Hawaiian deli in 1972 and, on a napkin, sketched the experiment that would create the biotech industry. Restriction enzymes (Werner Arber, Nobel 1978) cut DNA at specific sequences. DNA ligase rejoins the pieces. Plasmids replicate in bacteria. Combine them and you can clone arbitrary DNA at scale.

In 1976, Boyer and venture capitalist Robert Swanson founded Genentech. By 1982 the company shipped the first recombinant pharmaceutical: human insulin (Humulin), produced in E. coli.

"Boys, this is it. We've been waiting all our careers for this." — Paul Berg, on hearing Cohen and Boyer's plan.

The polymerase chain reaction is conceptually simple: heat to melt the strands, cool with primers and a thermostable polymerase, anneal, extend. Each cycle doubles the target. Thirty cycles yield a billion copies. Mullis used Thermus aquaticus Taq polymerase, isolated from a Yellowstone hot spring.

denature  → 95°C, 30 s
anneal    → 55°C, 30 s
extend    → 72°C, 60 s
× 30 cycles ⇒ 2³⁰ ≈ 10⁹ amplification

PCR underpins forensic DNA, COVID diagnostics, and most modern molecular biology. Mullis received the Nobel in 1993.

Frederick Sanger's dideoxy method (1977) used chain-terminating nucleotides to enumerate a sequence base by base. Throughput grew slowly until parallelized fluorescent capillary electrophoresis (Applied Biosystems, 1990s).

The next-generation revolution arrived with Solexa/Illumina around 2007: massively parallel short-read sequencing on flow cells. Long-read platforms (PacBio SMRT, 2010; Oxford Nanopore, 2014) closed the gap on repeats.

The Human Genome Project ran 1990–2003 across labs in six countries. Celera Genomics, founded by Craig Venter in 1998, used whole-genome shotgun sequencing to publish a parallel draft. The two teams jointly announced the draft at the White House in June 2000; the finished sequence followed in April 2003. The first complete telomere-to-telomere assembly came in 2022 (T2T-CHM13).

Clustered Regularly Interspaced Short Palindromic Repeats are the bacterial defense memory. Cas9 is a guided endonuclease: a 20-nt RNA programs it to cut a complementary DNA target. Jennifer Doudna and Emmanuelle Charpentier's June 2012 Science paper repurposed it as a programmable editor; Feng Zhang's lab at the Broad demonstrated mammalian editing months later. Nobel Prize 2020 to Doudna and Charpentier.

By 2023, the FDA approved Casgevy, a CRISPR-based therapy for sickle cell disease — the first authorized human gene-editing treatment. Base editors and prime editors (Liu lab) further reduced collateral DNA damage.

"You can imagine that any process that involves DNA could be edited." — Jennifer Doudna, 2015.

Katalin Karikó and Drew Weissman's 2005 paper showed that pseudouridine-modified mRNA could evade innate immunity. Building on that work, Moderna (founded 2010) and BioNTech (2008) developed lipid-nanoparticle delivery. When SARS-CoV-2 was sequenced in January 2020, the BioNTech-Pfizer vaccine candidate was designed within days; clinical efficacy was reported in November 2020.

DeepMind's AlphaFold 2 (2020) and AlphaFold 3 (2024) effectively solved single-domain protein structure prediction; ESM-2 and Boltz-2 brought open-weights versions. Generative models like RFdiffusion and Chroma now design de novo proteins with target binding sites. Twist Bioscience and Ansa Biotechnologies push synthetic DNA cost below $0.05 per base pair.

The result is a tightening loop: read, write, simulate, predict — and now generate.