One gene controls one trait, and mapping the human genome would explain most diseases.

In biology classrooms of the 1980s and 1990s, the gene occupied a clean conceptual position. You inherited genes from your parents; genes controlled traits; knowing someone's genes meant knowing something definitive about what they would become. The framework had deep roots. The one-gene, one-enzyme hypothesis, first articulated by George Beadle and Edward Tatum in 1941 through experiments with the bread mold Neurospora crassa, was later generalized into one gene, one polypeptide, and then simply one gene, one trait. Research biologists had already complicated this picture substantially. In textbooks and classrooms, the complication had not arrived.

On June 26, 2000, Francis Collins of the National Human Genome Research Institute and Craig Venter of Celera Genomics stood in the East Room of the White House beside President Clinton to announce a working draft of the complete human genome sequence. Clinton called it "the most important, most wondrous map ever produced by humankind." Tony Blair joined by satellite link from Downing Street. The Human Genome Project had formally launched in 1990 with the expectation that once the full sequence was in hand, the mechanisms of disease, inheritance, and human variation would come clear. For a decade, journalists wrote about finding the gene for intelligence, the gene for depression, the gene for cancer. The gene-for construction pervaded popular discourse and filtered into classrooms as the promise of what genomic science would soon deliver.

The completed sequence, officially declared finished on April 14, 2003, immediately complicated everything. The human genome contained somewhere between 20,000 and 25,000 protein-coding genes, far fewer than predictions of 80,000 to 140,000 — roughly the same number as a mouse, and only twice that of a nematode worm with fewer than a thousand body cells. More striking, coding sequences amounted to less than two percent of the total genome. The other ninety-eight percent, long dismissed in the scientific literature as "junk DNA," a term coined by geneticist Susumu Ohno in 1972, sat there unexplained.

The one-gene-one-trait model collapsed against these findings. Height is influenced by hundreds of genetic variants, each contributing a small fraction of the total effect. The same was true of intelligence, temperament, and susceptibility to most common diseases: type 2 diabetes, schizophrenia, coronary artery disease, most common cancers. The genome-wide association studies that followed found hundreds of disease-associated variants, nearly all of them of individually tiny effect, scattered across regions that often had no known protein-coding function. The hoped-for single-gene explanations materialized for only a small number of conditions.

The "junk" was the story. Beginning in 2003 and culminating in the ENCODE Project Consortium's landmark 2012 publication — thirty simultaneous papers coordinated across 440 researchers — investigators demonstrated that approximately 80 percent of the genome is biochemically active in at least one cell type. The non-coding regions contain regulatory switches (enhancers and silencers), RNA genes, binding sites for transcription factors, and structural elements that determine how the genome folds in three dimensions inside the nucleus. Mutations in these regulatory regions turned out to cause disease as readily as mutations in coding genes. The genome was not a parts list. It was a wiring diagram, and the wiring turned out to be almost the whole thing.

Epigenetics added yet another layer. Chemical modifications to the DNA molecule itself — methylation of cytosine residues, modifications to the histone proteins around which DNA is coiled — alter which genes are expressed in which tissues at which times, without changing the underlying sequence. These patterns are partly heritable and partly responsive to environment, meaning that two individuals with identical DNA could develop measurably different patterns of gene expression depending on their early life conditions.

The idea that sequencing a genome would reveal the person in the way a blueprint reveals a building — a metaphor that circulated widely in popular accounts of the project — turned out to misrepresent the relationship between sequence and organism. DNA provides possibilities and constraints. It does not, by itself, determine outcomes. What the genome project gave biologists was not a simple instruction manual. It gave them a new set of questions, most of which could not even have been formulated before the sequence existed.