Drug companies spend years and fortunes guessing which proteins are safe to switch off. Nature, in effect, has already run the experiment. Somewhere in every population there are people born with both copies of a gene broken, and their health tells you what happens when that gene stops working. The trick is finding them. A study published this week in Nature describes a biobank built to do exactly that, and it is anchored in a country that rarely shows up in large genetics datasets: Pakistan.
The Pakistan Genome Resource enrolled 173,303 participants and sequenced their exomes and genomes. Across that group, the team catalogued people carrying naturally occurring homozygous loss-of-function variants in 6,476 genes. Homozygous means both inherited copies are hit. Loss-of-function means the gene is effectively silenced. In genetics slang, these participants are "human knockouts", living versions of the engineered mice labs have relied on for decades.
Why relatedness is an asset here
Rare variants are, by definition, hard to study. You need enormous samples before a single broken gene shows up in two copies in the same person. Populations where marriage between relatives is common shift those odds. When parents share ancestry, a rare variant carried on both sides can land in a child twice, producing a full knockout that would almost never appear elsewhere. Pakistan has high familial relatedness, and the researchers leaned into that. The result is a catalogue of double-hit variants far richer than what a randomly mixed population of the same size would yield.
The authors did not stop at counting variants. They mapped how these loss-of-function events spread across molecular pathways, tied specific genes to blood biomarkers, and ran what they call recall-by-genotype studies. That last piece matters. It means going back to individuals with a particular knockout and examining them closely, treating a person's own genetics as a natural stand-in for a drug that blocks the same protein. If people missing a gene are healthy, blocking its protein with a medicine looks safer. If they show problems, that is a warning worth having before a trial, not after.
What a knockout can and cannot tell you
The logic is appealing because it sidesteps a stubborn problem. Animal models often fail to predict human biology. A person walking around without a functioning copy of a gene is the most direct evidence you can get. That is why the field has chased these variants for years, and why a South Asian resource of this scale fills a real gap. Most large genetic databases skew heavily European. A catalogue built in Pakistan adds variants and biology that those databases simply miss.
Worth keeping the claims in proportion. A knockout in a biobank shows association, not a guaranteed drug outcome. A gene that looks harmless when silenced from birth may behave differently when a drug shuts it down in a 60-year-old with existing disease. Timing, dose, and compensation over a lifetime all complicate the read. The paper presents a reference resource and a set of associations, not finished drug targets, and the authors frame it that way. Turning any single knockout into a therapy is still a long road with plenty of ways to fail.
There is also the question of who benefits. The biobank draws on Pakistani participants and a network of hospitals across the country, yet much of the analysis in the author list runs through pharmaceutical companies in the United States, Europe, and Japan. Building genetic resources in understudied populations is overdue. Making sure those populations share in the medicines and knowledge that follow is the harder, longer commitment.
A wider map of human genetics
Step back and the significance is less about any one gene than about the map itself. For too long, "the human genome" in practice meant a narrow slice of humanity. Adding 173,303 people from Pakistan, with their distinctive pattern of naturally silenced genes, widens what researchers can see. Some of those variants will be dead ends. A few might point toward the next generation of medicines. Either way, the reference now exists, and the people who contributed it are finally part of the picture.
Comments