The Gene Sequence

The result arrived, as the important ones often do, on an ordinary Tuesday afternoon, in the form of an image that nobody quite believed. Dr. Sofia Navarro had run her laboratory in Cambridge for seven years. Its work concerned regeneration: specifically, the remarkable ability of zebrafish, small striped fish no longer than a finger, to regrow a damaged fin within a couple of weeks. For over a decade, it had been widely accepted that a particular gene, which the lab referred to simply as R7, was essential to the process. Remove it, and the fin should not grow back.

The image on the screen showed a fish from which R7 had been removed. Its fin had grown back perfectly.

It was Tom Hadley, the lab's most senior postdoctoral researcher, who brought it to her, and it was clear from his face that he had already begun to imagine the paper. "If this holds up," he said, "it overturns the whole model." Sofia looked at the image for a long time. "If," she said. It was, her students joked, her favorite word. She had been known to circle it in red on other people's drafts, with a note in the margin: "Is it, though?"

The first thing to be done was to establish that the fish really lacked the gene. Samples were taken, the relevant region of DNA was sequenced, and the results were checked by two people independently. The deletion was confirmed. The experiment was then repeated with a fresh group of fish, bred from a different pair of parents, in case something about the original family had been unusual. Again, the fins regrew. By the end of the month, the effect had been observed in forty-three of forty-five fish. Tom had started drafting an abstract.

Sofia had not. Her reluctance was not, she insisted, a matter of temperament. A result that appears to contradict a well-established finding is, statistically speaking, more likely to reflect an error than a revolution. That is not because established findings are always right; it is because there are a great many ways for an experiment to go wrong and only a few ways for a long-standing theory to have been mistaken. "Extraordinary claims," she reminded the lab meeting, "require us to be extraordinarily boring first."

Tom found this maddening, and said so. A rival group in Boston was known to be working on the same gene, and if they published first, two years of work would be reduced to a footnote. He argued, not unreasonably, that the data were clean, the controls had been run and the sample was larger than most studies in the field. Science that waited for perfect certainty, he said, would never publish anything. There was some truth in that, and Sofia acknowledged it. But there was a difference, she replied, between waiting for certainty and refusing to look.

It was Priya Raman, a second-year PhD student who rarely spoke in lab meetings, who eventually made the difference. She came to Sofia's office one evening, closed the door, and said, very quietly, that she had noticed something weeks ago and had not mentioned it because she assumed she must be wrong. In the fish without R7, a second, closely related gene, a sort of molecular cousin that the lab had always regarded as unimportant, appeared to be far more active than normal. "It might be nothing," she said. "I didn't want to slow things down."

It was not nothing. Over the following weeks, it was shown that when R7 was deleted, its cousin was switched on at much higher levels, and appeared to take over much of its role. The phenomenon was not unknown; biologists have described similar cases of so-called genetic compensation in several organisms, in which the loss of one gene seems to be at least partly offset by the increased activity of a related one. When both genes were disrupted together, the fins did not regrow. The original model, it seemed, had not been overturned. It had been incomplete.

The disappointment in the lab was considerable, and Tom felt it more than anyone. For a few days he barely spoke. Then, to his credit, he came into Sofia's office and admitted that he had wanted the first result to be true so badly that he had stopped asking whether it was. "I was writing the press release in my head," he said. "I wasn't reading the data any more. I was reading my own CV." Sofia told him that every scientist she had ever respected had done the same thing at least once. The ones she respected most had done it and then gone back to the bench.

The paper that was eventually published was more modest than the one Tom had imagined, and, by most accounts, considerably more useful. Its title contained the word "compensation" rather than "overturns". Its conclusions were carefully hedged: the data were described as "consistent with" a partly redundant role for the two genes, and the authors noted that "further work will be needed" to establish whether the same mechanism operates in other tissues. Priya Raman was listed as joint first author, alongside Tom, at Sofia's insistence.

The group in Boston published a few months later. Their paper, which appeared in a more prestigious journal, reported that R7 was not essential for regeneration after all. It did not mention the second gene. Within a year, two other laboratories had reported that they could not reproduce the result unless the compensation was taken into account, and the Boston group issued a correction. Tom printed the correction out and pinned it, without comment, to the noticeboard by the coffee machine. Next to it, someone had written a single word in red ink and circled it: "If."