AS THEY GATHERED AT COLD SPRING HARBOR ON 12 MAY 1998 FOR the annual symposium on genome mapping and sequencing, the leaders of the Human Genome Project were in various stages of shock, anger and despair. The previous Friday, 8 May, word had gone out that Craig Venter had obtained private backing to form a commercial company with the declared intention of sequencing the entire human genome in three years. The as-yet unnamed company would become, said Craig’s publicists, ‘the definitive source of genomic and associated medical information that will be used by scientists to develop a better understanding of the biological processes in humans and deliver improved health care in the future.’ The press release, issued two days before the conference opened, talked of plans to ‘make sequencing data publicly available’ by releasing data every three months, but at the same time declared Craig’s belief that ‘this information has significant commercial value.’ I could see no other interpretation than that Craig was aiming to gain total control of the information contained in the genome for commercial gain. His whole philosophy appeared to run directly counter to everything we had fought so hard to achieve through the Bermuda agreements.

There had been almost no forewarning. Only a week before, Jim Watson heard what was up through a call from Richard Roberts. Roberts, though British-born, had spent much of his scientific career at Cold Spring Harbor working alongside Jim, and was co-discoverer of ‘split genes’, the phenomenon by which non-coding introns interrupt the coding sequences or exons. Since 1992 he had been based at a private company, New England Biolabs, which supplies restriction enzymes (many of which Roberts discovered) to the molecular biology community. Roberts told Watson that he was chairing the scientific advisory board of Craig’s new company, gave him an outline of what was planned and asked him if he wanted to be involved. ‘I wondered why Craig didn’t call me himself,’ says Watson, who was sufficiently worried to call Michael Morgan at the Wellcome Trust. Michael was crossing London in a taxi at the time, on the way to a planning meeting for a proposed expansion of the Genome Campus. As soon as he got the message, he rang Jim back from the taxi. ‘He told me that Craig was going to make a major announcement on Monday about sequencing the genome in a year, privately—what were we going to do about it?’ says Michael.

Michael had been working with us for the previous few months from the Trust’s side on a renewed submission of our proposal to ramp up from a sixth to a third of the genome, to match serious United States commitments expected later that year. His first thought was that Craig’s initiative would torpedo the proposal. ‘I thought the governors would say, “If it’s going to be done, wonderful—why should we put any more money into it?”’ says Michael.

At that stage all we knew was that Craig had got a lot of money from an industrial source. Then I heard from Bob Waterston that Francis Collins and the NIH director Harold Varmus had been invited to meet Craig on Friday 8 May. It was only at that meeting, at the United Airlines Red Carpet Club at Dulles Airport in Washington, that Collins found out that the commercial partner was Mike Hunkapiller of ABI, the company that made the sequencers

used by almost all the genome labs. Hunkapiller had been a key figure in Lee Hood’s CalTech lab, where the sequencers were originally developed, and between them they had launched ABI to make the machines commercially, with Hunkapiller at its head. In 1993 the company had been bought by the scientific instrument makers Perkin-Elmer, and it was Perkin-Elmer that was putting up the money, $300 million or more, to launch Craig’s new sequencing initiative. Its chief executive, Tony White, was in the process of turning a successful instrumentation company into one of the most aggressive and successful players in the high-technology sector. It had been Hunkapiller’s idea to invite Craig to head a new subsidiary specializing in selling genomic information and software tools for genomic analysis.

At first I didn’t fully realize the size of the challenge. I did see immediately that we were no longer going to be the largest genome center in the world. That was disturbing. Anyone who had more machines could dominate in terms of speed, not just on the human genome but on other sequencing projects. For example, after finishing yeast Bart Barrell’s team at the Sanger Centre had embarked on a series of projects to sequence pathogens that caused diseases such as tuberculosis and malaria. I remember telling Bart that we would have to be on our guard. He was already used to facing challenges from Craig Venter, having successfully raced TIGR to complete the sequence of the TB bacillus. I also saw that we might have to compete on the human sequence by adjusting our own strategy to do more upfront sequencing, rather than keeping the sequencing and finishing in step as we had up to now. This was, after all, the essence of Bob’s proposal of more than three years before. It wasn’t a matter of racing to be first and claiming the glory. If there was a danger that private interests could gain control of the fundamental information in the genome sequence, it would be more important than ever to push that sequence out as fast as possible.

The weekend before the Cold Spring Harbor symposium passed

in a flurry of e-mails as we at the Sanger Centre and our colleagues in the United States tried to assess what our response should be. The official line from Francis Collins’s office quickly emerged as cautiously welcoming. Reporting on the Friday meeting, Mark Guyer, who was Francis’s deputy at the National Human Genome Research Institute (NHGRI), wrote to all the genome center heads to say, ‘This is a very exciting development, providing a major infusion of resources and new technology at a critical juncture in the progress of the genome project.’ In the first of what was to become a series of bizarrely staged shows of unity, Craig and Mike Hunkapiller appeared alongside the NIH director Harold Varmus, the NHGRI director Francis Collins, and the director of the Department of Energy’s Office of Biological and Environmental Research, Ari Patrinos, at a press conference in Washington on the morning of Monday, May 11. Craig and Mike and some of their colleagues were to attend the Cold Spring Harbor symposium, which began later in the week, to ‘encourage discussion and interaction about their plan.’

I hadn’t planned to go to Cold Spring Harbor at all. I was hardly going to any meetings at that time—as far as I was concerned my most useful contribution was getting the worm genome finished, a goal that was now well within reach. Once the Sanger Centre was up and running, I had always delegated everything else: David Bentley was in charge of human genetics, Jane Rogers in charge of sequencing, Richard Durbin in charge of informatics, and so on. I took the view that they should be the ones to represent the Sanger at international meetings. Maybe that made me a rather unusual sort of director, but it emphasized the strength of the team. People were responsible for what they were doing, and I think that was recognized as a strength in the end. But after the news broke about Craig’s company, Bob and others urged me to go after all. The Venter plan was a potential threat to the whole genome project, and the heads of all the publicly funded genome labs needed to formulate a response as fast as possible.

I was due to meet the Wellcome Trust governors on the Wednesday of the symposium week. As arranged months before, we would be presenting the new bid to ramp up human sequencing. Michael Morgan and I decided at once that if we got the money we should go over to Cold Spring Harbor straight afterwards and make our own announcement.

Craig did not as a rule come to the Cold Spring Harbor symposia, but ran his own TIGR-sponsored events. On this occasion there was a ‘pre-meeting’ for principal investigators funded by NHGRI scheduled for the day before, which as leader of the TIGR group (TIGR was sequencing part of chromosome 16) he would be expected to attend. Whatever the original agenda, the priority was now to hear at first hand what Craig was proposing to do. He duly made his case to the heads of the other sequencing labs. Not surprisingly, the reception was hostile. Not only was he going into direct competition with his former collaborators, but he had already secured a massive public relations advantage in giving the whole story to an appreciative New York Times journalist, Nicholas Wade, for the May 10 Sunday paper. Anyone reading that article who was not familiar with the complexities of genome sequencing would have no reason to doubt that Craig was about to pull off an extraordinary scientific feat that would leave the publicly funded project standing. At the Washington press conference the day after the article came out, he had suggested that perhaps it would be better for everyone if the HGP left the human genome to him and turned its attention to sequencing the genome of the laboratory mouse. He repeated this proposal to the stunned researchers at Cold Spring Harbor. ‘It was like asking them to walk into the sea and drown,’ says Jim Watson, who although no longer responsible for the project retained an understandably proprietorial interest. ‘It is an understatement to say it was done in an insulting fashion.’

At his meetings with Varmus and Collins, Craig had apparently proposed that NIH collaborate with him by sharing data. If he really

meant this to be a two-way deal, it might have been interesting—of course, all the HGP data was publicly available anyway. But at the same time, as the press conference and New York Times article made clear, he was implying that as far as the human sequence was concerned, the public project would become redundant. ‘Congress might ask why it should continue to finance the human genome project through the National Institutes of Health and the Department of Energy,’ Wade had written, ‘if the new company is going to finish first.’

The anguish at Cold Spring Harbor was palpable. Jim Watson was so incensed that, in characteristically unguarded language, he compared Craig’s attempt to take over human sequencing with Germany’s invasion of Poland, and demanded to know whether Francis Collins was going to be Churchill or Chamberlain. He feared, for a few days at least, that the network of international collaboration that he had done so much to foster might be destroyed overnight. ‘I worried that there would be the perception that we couldn’t win,’ he says.

That worry was certainly justified. On the face of it, Craig’s proposal looked extremely strong. He himself was a scientist with a proven track record in running high-throughput sequencing labs. His commercial partner was the head of the company that made the sequencing machines, ABI. The joint venture was funded to a degree that was beyond the wildest dreams of any of the individual genome project labs. Large-scale genome sequencing is expensive: huge amounts of money are needed to buy the machines, hire the sequencing teams, and above all pay for a constant supply of expensive reagents—the enzymes and nucleotides used in the sequencing reactions. And the venture was founded on an important technological advance. ABI had now prototyped its new sequencing machine, the 3700, which could produce sequence several times faster than the previous model. Instead of running the DNA samples through lanes in slabs of gel, the new sequencers ran

each sample through a fine gel-filled tube, or capillary, which made it possible to automate the loading of samples and reduce the amount of human intervention needed. The new company was to be endowed with 230 of these machines, each capable of reading 96 samples simultaneously. It aimed to have an output of 100 megabases of raw sequence per day.

There was absolutely no doubt that, armed with these resources, Craig was capable of equalling or even exceeding the existing world output of raw genome sequence, assuming that the as yet untested machines performed as they should. But as Bob and Eric and the other genome leaders questioned him about the details of his strategy, they became convinced that it just wouldn’t work. Craig was proposing to speed the process up by using the whole-genome shotgun method he had employed to sequence bacteria such as Haemophilus influenzae and Helicobacter pylori. Instead of mapping clones by fingerprinting and shotgun sequencing them one by one, as we had done for the worm and were doing for the human, he was going to shotgun the whole genome at once.

It wasn’t the first time someone had proposed to do this. At the Bermuda meeting in February 1996 James Weber, a human geneticist from the Marshfield Medical Research Foundation in Wisconsin, argued that the search for human disease genes could be greatly speeded up by the whole-genome shotgun approach. In partnership with Gene Myers, a computer specialist from the University of Arizona in Tucson, he followed up the presentation with an article in the journal Genome Research. But the genome project scientists rejected the idea. In a closely argued response to their article, Phil Green of the University of Washington, who knew as much as anyone about the problems of assembling human sequence, pointed out a number of shortcomings with the proposal. The most serious, from a scientific point of view, was that the human genome was a very different proposition from the bacterial genomes on which the strategy had previously succeeded.

Bacterial genomes consist of about 2 million bases and the human genome 3 billion. But they’re also qualitatively different, especially in terms of the amount of repetition there is in the genome. Bacterial DNA typically includes about 2 percent repeat sequences; human DNA includes about 50 percent. Repeat sequences mean that one DNA fragment might have matching overlaps with many other fragments. Only by working with fragments taken from larger clones whose position in a whole-genome map is already known is it possible to be sure you’ve assembled the whole sequence correctly. Furthermore, some regions of DNA are much harder to clone in bacteria than others, so a random approach leaves many more gaps than you might predict on a purely theoretical basis. One of the reasons Bart Barrell’s group was able to complete the TB sequence ahead of TIGR was that the whole-genome shotgun approach had not worked in that case. TB is exceptional among bacteria in having a lot of repeats that are also hard to clone.

Craig might have managed to gather enough computer power and software expertise to solve the assembly problem through sheer brute force. But we wouldn’t know until it was too late. Our method, mapping first and then sequencing the mapped clones, though laborious at first sight, meant that you could check how good your data were as you went along. The quality of the product was guaranteed, and our priority had always been to produce a sequence that would stand for all time. There was no doubt that the whole-genome shotgun would deliver plenty of raw sequence, but how many pieces would the assembly end up in? If it turned out to be full of holes, it would have some use for gene hunting; but it would be prohibitively expensive to finish, and the location of the smaller pieces on the genome would be unknown. Phil Green had made all these points in print the year before, and yet Craig was going ahead anyway.

But the greatest concern was over exactly what he intended to do about releasing his data. Officially they were to be released ‘on a

quarterly basis.’ But it was difficult to see how the company could succeed as a commercial supplier of genomic information if it did indeed make all its data public in this way. The initial announcement said that it would be seeking patent protection on ‘only 200–300 genes’—not a very competitive position when other companies, such as Incyte and Human Genome Sciences, were already claiming rights to many more. So while the HGP leaders were doubtful about Craig’s science, they were even more doubtful about his business plan. It would be all too easy for him to backtrack on his early commitments in order to satisfy his shareholders. ‘It was a testy meeting,’ says Bob Waterston. ‘We asked Craig hard questions, and he thought we were just being nasty to him. And we might have been a little nasty, but scientists are skeptics. Craig has always been very thin-skinned that way—he reacts very badly to criticism.’

Craig had one answer to critics of his scientific strategy: he announced that he was going to test the method on a smaller organism first, and then, within the hearing of many of the conference participants, asked Gerry Rubin to step outside and have a word. The team Gerry had assembled in Berkeley, California had begun to sequence the fly genome in 1995. By 1998 they had done 20 megabases out of 120, and were planning to scale up to finish the rest, with the help of Richard Gibbs who now had a large sequencing center at Baylor College of Medicine in Houston. As they saw the burst of discoveries that followed our steadily increasing output of the worm sequence, the much larger fly research community was clamoring for the fly genome to be completed faster.

Gerry just wanted to get the fly done—he had already put more time into it than he wanted to, having taken it on in the first place largely because nobody else looked like doing it. So when Craig told him that his company was going to sequence the fly and would like to do it as a collaboration, he was prepared to listen.

Gerry is a pragmatist; he saw immediately that he could not compete with Craig and that it would be to his advantage to throw in his lot with him. I talked to him about what had happened and agreed that he had no choice. A few months later Celera—as Craig’s company was eventually named—formally signed a memorandum of understanding with the University of California (the official recipient of the NIH grant which funded Gerry’s sequencing work), including the commitment to release the data freely.

It was surprising to a number of people that Gerry agreed so readily. Scientists in the academic world stake claims to little plots in the field of scientific inquiry which others by and large respect. Although there are plenty of examples of ‘races’ to make new discoveries and arguments over priority in the history of science, the processes of grant distribution and publication make them much less frequent than one might expect. True, if someone doesn’t seem to be getting anywhere then others might move in and take over. But no publicly funded scientist, in the United States at least, would have applied for a grant to sequence the fly genome because everyone knew that Gerry was doing it. The NIH would not have funded such a grant anyway because funding bodies don’t like duplication. Part of the reason why many people find Craig hard to stomach, and why others admire him greatly, is his cavalier disregard for such academic niceties—and with Perkin-Elmer’s money behind him, he could afford it. But Gerry says he had never felt possessive about the fly genome project; his little plot was in fly molecular genetics, and he wanted to get back to learning how flies are put together as soon as possible. He kept a small genomics program going in his lab in case Craig’s effort failed, but otherwise agreed to hand responsibility

for the shotgun part of the sequencing effort over to the company.

Craig was too late to take control of the worm. Back in 1990 labs had been taking bets about which would win, fly or worm, and at that time we were not the favorites. But once Bob Waterston and I got going, the worm opened up a huge lead; by the time of Craig’s announcement it was almost done. If we had not been so far ahead, he might just as easily have picked on the worm; or if he had not, the fly would certainly have caught the worm up once he took it on. It was good that, at Bob’s insistence, we had gone faster with the worm sequence than was originally planned. If we had gone at my pace and not Bob’s, Craig would have been able to catch us up. Why did it matter to be first? It’s not just a question of personal pride. Our labs by this time were large-scale enterprises, employing a lot of people and with a great deal of hard-won money invested in them. We had to set ourselves tough goals and meet them in the face of competition in order to justify the investment that had been made in us, and keep the credibility that would ensure our future funding. These are the forces that push science forward efficiently. In addition, our scientific credibility gave us influence over the quality of the product and how data were handled.

Craig had declared his intention of invading the fly genome and looked like taking the flag without a fight. Would he be able to do the same with the human? He had certainly opened up a huge public relations advantage, something that was to happen time and again in the years to come. I’m sure he has good advisers, but there’s no question that his own instinct for the timing and presentation of announcements, and for the apparently casual aside, is superb. Again, it’s a skill that most scientists never learn and rarely have to practice. The convention is that you don’t announce anything until the work is completed and accepted for publication in a peer-reviewed journal. That way there should be no embarrassing retractions after hasty announcements, as happened in the case of Stanley Pons and Martin Fleischmann’s ‘discovery’ of cold fusion in

1989. But Craig was no longer in science, he was in business. And the priority for a business is not scientific credibility but share price and market penetration.

The initial press reports on Craig’s company, beginning with Nicholas Wade’s New York Times article, set a trend for presenting the new venture as a more effective alternative to the Human Genome Project. Just to take two examples, it was said that the new company would sequence the genome faster, and would do it more cheaply. The initial Perkin-Elmer press release said that the company would have a ‘substantially complete’ genome in 2001, four years ahead of the Human Genome Project’s target of 2005. It did not mention cost, but most articles quoted the figure of around one tenth of the $3 billion budget of the HGP suggested by Craig at the 11 May press conference. But these reports weren’t comparing like with like. The public project’s target date was for finished sequence, with all possible gaps closed and to an accuracy of 99.99 percent. Craig’s company, as he made clear to the genome center leaders at Cold Spring Harbor, did not plan to bother with the time-consuming process of closing gaps, which takes a lot of manual intervention, but aimed to get as much as he could out of a fully automated procedure and then stop. Although Craig later insisted that his sequence would be ‘highly accurate’ and ‘comparable to the standard now used in the genome sequencing community of fewer than one error in 10,000 base pairs’, it was clear (as subsequently turned out to be the case) that the whole-genome shotgun method would produce what we now call a draft, not a fully finished sequence.

As for the cost, the Human Genome Project funded far more than human sequencing. Apart from other genomes, such as worm, fly and yeast, it had a program on the ethical, legal and social impact of genome research and a number of projects on the development of new technologies, and supported databases and bioinformatics. Craig also said at the press conference that he could get the cost of

sequencing down from the genome centers’ then figure of 50 cents to 10 cents a base. But that cost, which the best centers were already undercutting, was for finished sequence, and it’s the finishing stage that accounts for at least half the cost: on this basis, the target of 10 cents was the one Bob Waterston and I had suggested two years earlier. Another widely quoted criticism of the HGP was that halfway into its fifteen-year schedule it had sequenced only 3 percent of the human genome; but we had deliberately spent the first six years scaling up through smaller genomes and developing the techniques. As the HGP’s founders had planned in the 1980s, the whole field had moved forward to the point where tackling the human genome was feasible, when earlier it had not been. Indeed, our purchases from ABI, and collaborations with them, had helped to drive the development of the faster sequencing machines that Craig’s company was planning to use—and which we too had every intention of using. Finally, that figure of 3 percent represented only finished sequence. Taking into account the assembled but unfinished clones that were in the databases, the figure was 13 percent.

As head of the National Human Genome Research Institute, Francis Collins had to take the brunt of the overt and implied criticism of the genome project. At the Sanger Centre we had the luxury of knowing that we had a politically independent body, the Wellcome Trust, behind us all the way. But in the United States, with a Republican Congress, Francis feared that any suggestion that the private sector could do a better job than government-funded labs was going to find a receptive audience.

It was also potentially a serious threat to the very survival of the publicly funded project in the United States.

Trying to get reporters to print the admittedly more complex analyzes that we felt were being ignored was going to be an uphill struggle. We were learning fast that we would have to play the public relations game if we were to survive. But that didn’t mean indulging in empty hype. What we needed was a big vote of confidence in the public project as a counter to Craig’s hints that it was an expensive white elephant. And that’s exactly what Michael Morgan and I hoped we would get from the Wellcome Trust governors that Wednesday. Long before we had any hint of Craig’s company, we had carefully prepared our case for doubling our output. Jane, David and Richard had all put together detailed accounts of how it would be done, with breakdowns of the cost—all the stuff that I can’t handle. I went up to London with Jane and Richard in the train. In my bag I had one handwritten overhead, containing a few bullet points; I scrawled an extra one about Craig’s company at the bottom.

Seated round the table in the panelled boardroom at the Trust’s offices in Euston Road on May 13, the scientific governors of the Wellcome Trust waited expectantly. I spoke for only a few minutes. I told them that I thought we had a very strong case, and that the U.K. could not continue to play a leading role in the Human Genome Project unless we took on more than one-sixth of the work. Then I added that the arrival of Craig’s company made it crucial that we should maintain a strong presence, if genomic information were to be made freely available to all. To the question ‘Why should we do it?’ I had a ready answer. Three years before it had looked as if other countries outside the United States, particularly France, Germany and Japan, were prepared to take a serious stake in the project. But by 1998 it was clear that this was not going to happen; they would only ever be minor players. If there was going to be an influential international stake at all, it would have to be taken by the

U.K. And without a strong international presence, there was a real danger that the whole project would fall into the hands of the commercial sector.

We went outside to wait for their decision, and soon afterwards the message came back that they had agreed: our funding would be doubled, giving us the resources to take on and finish a third of the human genome. Once the governors understood that Craig Venter’s initiative was essentially a privatization of the genome, Michael Morgan felt there was no risk that they would pull out on us.

It was an emotional moment. I sent a message back in to the governors thanking them for their support, and saying that I saw us all as joining together in a partnership to meet our sequencing goals. What I didn’t say was that with the new grant we felt the governors had shown their trust in us to an extent that we had often wished for but not always experienced in the past. Yes, they had invested hugely in us, but somehow there had always been a feeling that they didn’t quite believe in us. The loss of the MRC’s support had made things very difficult, and maybe no more could have been done anyway, but I still felt the Trust had missed an opportunity in not giving us the green light to go for a third of the genome in 1995. For the next two years there was a feeling among us that the Sanger Centre was just not doing what it was designed to do. We did a lot of things, and we were successful, but we were forced to mark time to some extent. We offset this as much as possible, though, by scaling up where we could and committing resources in readiness. Richard Durbin says now that I drove the scale-up very strongly, claiming extra chromosomes, moving worm sequencers across to human as the worm neared completion, and ‘engaged in justifiable brinkmanship.’ I

don’t recall doing much driving, but I certainly made no attempt to rein it in.

With the new grant, all that frustration could be put behind us. Michael was full of excitement—the first thing he did was open a bottle of champagne, and then he and I immediately sat down in the press office with Jane and Richard and drafted a press statement. Jane flew to New York that evening, and Michael and I followed the next day.

It was difficult to sustain our euphoria that first evening at Cold Spring Harbor. Everyone else was very down. Craig had left before the symposium proper even began, tight-lipped at the criticism he had received but in no mood to compromise. What was needed was a robust, coherent response, but there was little agreement about what that response should be. Broadly speaking, people had divided into two camps. One side (most emphatically represented by Eric, who on this occasion was right) argued that we should immediately change our strategy and speed up the production of sequence to beat Craig at his own game, and worry about finishing later. The other (including a number of those in the Sanger Centre and Bob’s lab in St. Louis) favored the view that we should not be bounced into such a major change of direction, but stick to our existing strategy of sequencing and finishing in step. There was pretty much agreement on two points: that however we got there, our ultimate goal was still to produce 99.99 percent accurate finished sequence, and that, as far as the American side was concerned, we stood in real danger of losing everything if, as we feared from the tone of his press statements, Craig were to persuade Congress that it was wasting its money.

Weary and jet-lagged, I sat with Bob and his group from St. Louis in the Blackford restaurant, anxiously going over our dilemma. The group was strongly against a change in strategy, but feared that going at the same pace would allow Craig to make political capital out of the HGP’s ‘slow progress.’ Rick Wilson said gloomily, ‘This is

a lose—lose situation.’ Earlier, almost as soon as I arrived, Jane had grabbed me, saying, ‘You’ve got to do something!’ There was a real possibility that the public sequencing effort in the United States would collapse, but even if it wasn’t that bad she feared that our tried and tested map-based methods might be under threat if those of our own HGP partners such as Eric Lander, who had less to lose from a shift to a less systematic approach, won the day. Eric seemed more upset than anyone about Craig’s announcement, and at one point he had a furious argument with David about mapping. I found all this very puzzling, because at this point he didn’t appear to have much of a stake in the outcome. It was not until much later that I finally understood that he was positioning the Genome Center at the Whitehead Institute to be the leading producer of raw sequence data.

Throughout that first evening I caught grins on the faces of symposium participants who were not involved in large-scale sequencing as they saw the members of the genome labs huddled grimly in corners. It was quite clear that they were enjoying the thought that this very small group, who had taken so much in funding to produce the human sequence, were finally getting their come-uppance. And even those who weren’t laughing did not really understand what we were up against. Mike Smith, who headed a genome center in Vancouver, wandered into Bob’s meeting obviously expecting to join in with a bit of academic debate as one normally would in the breaks at the symposium. I told him it was a private meeting, and he was clearly upset. I sat with him at breakfast the next day and talked it over. The point is that the sequencers were no longer running traditionally structured labs, with a group of more or less independent scientists and a few technicians in support; we were effectively running ‘businesses.’ Bob and I had the biggest businesses at the time; Eric Lander aspired to have the biggest business. It was not just a matter of being on your own and so able to chuck everything out of the freezer and start again if a line of

research didn’t work out—which is the way you should do science. We had got ourselves in a position where we had highly trained staff now numbering hundreds. You can’t treat an organization like that as a personal research group. It looked as if we were just being pig-headed and defending our interests, but we were in a position of responsibility in more ways than one: without us, the human genome would be privatized.

I tried to explain this to Mike Smith, but I don’t think he understood. ‘You’re not leaving anything for anybody else,’ he protested. But if the project had been left to all the small groups, Craig would have just walked all over them. We were facing huge pressures, and the sense of what was right and what was wrong was quite different from the normal senses of right and wrong that you might come across at biological meetings. Mike, who tragically died of cancer in October 2000, was a wonderful man, but Bob and I had been forced to adapt from his way of thinking in order to make things happen. And at that time the most important thing was to maintain morale in the lab, talk through what the Venter announcement meant for us and think about strategy.

My feeling at the time was that we didn’t need a radical change of strategy, but that we should scale up and increase efficiency. And unlike the Americans, we at the Sanger Centre knew that we had the money to do it. The timing could not have been better if the script had been written in Hollywood. On the Friday morning, the symposium organizers gave Michael Morgan and me a slot before the scheduled program to make our announcement. Grace Auditorium, the lecture theatre on the Cold Spring Harbor campus, was packed. Michael got up to speak first. He explained that, in funding the Sanger Centre to double its output, the Wellcome Trust was reaffirming its commitment to an international initiative to produce a high-quality, finished product that would be publicly available for all to use. Alluding to the Venter—Hunkapiller proposal to patent some sequences, he threw out a challenge: the Trust was

opposed to the patenting of basic genomic information, and would be prepared to contest such patent applications in the courts. It was just what everyone needed to hear. When he had finished, the crowd in the packed hall rose to its feet in acclamation. Michael has a great sense of theatre and knew exactly what he was doing: putting pressure on the National Institutes of Health to increase its backing for genome sequencing. He really enjoyed the moment. ‘The atmosphere was electric,’ he says. ‘It was fantastic, very exciting.’

I stepped up straight afterwards to add a few words. I suddenly caught sight of Jim Watson at the side of the auditorium, his eyes glinting in the half-light. It immediately brought back to mind a scene in the BBC film Life Story, about the discovery of DNA. In a beautifully lit shot Jeff Goldblum as Watson is at the back of an auditorium watching a lecture on DNA. Knowing Jim was there, I ended with the slightly trite but nevertheless appropriate comment that whatever happened, it was now more rather than less likely that the sequence would be completed in the lifetime of a certain individual. And the crowd erupted again, stamping their feet in approval.

It was a very significant development indeed. Francis Collins called Wellcome’s decision ‘a shot in the arm’, but Jim Watson goes even further: ‘It was absolutely critical, psychologically,’ he says. In Bob Waterston’s view it made it possible for NIH, which had initially been too timid in its response, to ‘stand up a little straighter.’ After that I heard less defeatist talk about Congress pulling the plug on the HGP. The next editorial in Nature reported that ‘The talk was of healthy competition rather than throwing in the towel.’ The question was, how should we proceed? Craig’s sudden appearance on the scene forced us to look for a strategy that would get visible results quickly without compromising ultimate standards.

The first opportunity for the HGP leaders to put their heads together came when we gathered at Airlie House in Virginia,

an NIH conference center, at the end of May. The purpose of the meeting was to sketch out the next five-year plan for the project, covering the years 1998–2003. Richard Durbin and I both went to represent the Sanger Centre, and Michael Morgan was there too for the Wellcome Trust. Mike Hunkapiller was there representing ABI and, by implication, Craig’s company. It was a tense meeting. Harold Varmus made it clear that there could be a lot of extra money available for the United States centers, but that grants would depend on an agreed strategy. It was obvious that the strategy would involve some form of acceleration, and some labs were in a much better position than others to rise to this challenge. And there were still important differences of opinion about whether we should accept the pressure to go for a fast intermediate product—a draft—or whether we should continue on our steady course to the finished sequence.

By the end of the meeting there was more or less an agreement, though with strong dissenting voices (mainly those of Richard Durbin and Phil Green), to try to get rapid shotgun coverage of mapped clones covering the whole genome—perhaps threefold coverage rather than the sixfold that was usual for the shotgun stage—at the same time as accelerating our production of finished sequence. Another objective I had was to make people realize that we had to take the threat of Craig Venter’s PR skills seriously, and actively promote the public project. Towards the end of the meeting, one balmy evening by the pool, Bob and I cornered Francis Collins, Michael Morgan and Alan Williamson and won their verbal agreement (as I thought) to run a publicity campaign emphasizing the positive points of the public project and correcting disinformation from the other side.

Bob and I were both fairly comfortable with the decision to go for a fast shotgun, or draft. Since the days of the worm, we had always been accustomed to the idea of releasing unfinished data. But we both met opposition within our own labs—in my case, Jane and

particularly Richard were unhappy about the idea of decoupling sequencing from finishing. And there were still major unanswered questions about how the whole job would be shared out among the partners in the international consortium.

Before the project’s leaders reached a final decision the United States Congress, or at least one of its subcommittees, called a hearing to investigate how the launch of Craig’s still unnamed company might affect the federally funded Human Genome Project. So, on 17 June, Craig lined up to address the Subcommittee on Energy and Environment of the House of Representatives Committee on Science, along with Ari Patrinos of the Department of Energy, Francis Collins from the National Human Genome Research Institute and Maynard Olson from the University of Washington.

Maynard has in many ways acted as the conscience of the genome project in terms of ensuring the quality of the final product. He was one of those who publicly doubted the wisdom of the ‘megalomaniac genome project’ when Bob and I first suggested it (see chapter 4), on the grounds that it might reduce the chances of a fully finished sequence ever being produced—although ultimately he supported us. Not one to rush into unconsidered judgements, his view of Craig’s announcement, read out to the assembled members of Congress, was damning.

He went on to issue the standard academic challenge to a doubtful claim: ‘Show me the data.’ And he predicted that Craig’s whole-genome shotgun method would encounter ‘catastrophic problems’ at the assembly stage, with 100,000 ‘serious gaps’ in the sequence. (Three years later he turned out to be spot-on.)

Francis echoed this criticism in his own presentation, and made an unfavorable comparison between Craig’s stated policy of releasing his data quarterly and the daily release policy of the public project. ‘Any delay can result in wasted effort in research,’ he said. He also pointed out, presciently as it turned out, that ‘the private effort’s commitment to data release might diminish over time, if business pressures came to the forefront.’ But he made it clear that ‘the private and public genome sequencing efforts should not be seen as engaged in a race,’ and that both his institute and the Department of Energy would welcome the opportunity to collaborate with Craig’s company.

When it came to Craig’s turn, he gave the committee a quick resumé of his past scientific record, including the claim that he and his colleagues at the NIH had ‘developed a new strategy for identifying genes’ (something of an overstatement as we saw on p. 108) before going out of his way to praise the achievements of the HGP. Moving on to talk about his company, he first assured them that ‘an essential feature of the new company’s business plan is to provide public availability of the sequence data.’ Addressing concerns about patents, he went on to make the claim that ‘Our actions will make the human genome unpatentable,’ but added that his company would seek intellectual property protection on ‘fully characterized important structures’ amounting to 100–300 targets. Arriving finally at the issue the committee was meeting to address, he said that the impact of his company on the HGP should be ‘to reorient it sooner to move beyond DNA sequencing into the research that will help us better understand and treat…diseases.’ He also dismissed the idea of a ‘race’, saying that the public project should be judged ‘by its ability to adapt and work with new initiatives rather than compete against them.’

I found it impossible to accept Craig’s words at face value, but somehow he emerged with his credibility intact. Maynard Olson’s clear-eyed criticisms, all of which turned out to be accurate, were

interpreted by many commentators as sour grapes. On the other hand, nothing the congressional committee heard that day appeared to persuade its members that it was time to stop supporting the work of the NIH and DOE on human genome sequencing. Francis had been able to show that, far from dragging its feet, the public effort had met all its targets to date on time and within budget, and that it expected to meet its new targets just as efficiently.

After three months in which we had to refer to Craig’s company as ‘the Venter-Hunkapiller proposal’ (usually shortened to Venterpiller or VentiPEde), it finally found a name, Celera Genomics Corporation, and a catchy slogan: ‘Speed matters. Discovery can’t wait.’ The company set itself up in Rockville, Maryland, next door to TIGR’s laboratories, in two block-like buildings, one for the computer center and one to fill with the new 3700 sequencers once they started to arrive. It was time for Francis Collins to seize the moment and make a statement as bold as those already made by Craig and by the Wellcome Trust. After much negotiating, he finally came out with a statement of research goals for the next five years in mid-September. The main points were that the international consortium would produce a ‘working draft’ sequence covering more than 90 percent of the total at an accuracy of more than 99 percent by 2001; and that we would go on to complete the fully finished sequence, with fewer than one error in 10,000 bases and no gaps, in 2003, fifty years after the discovery of the DNA structure. ‘No-one else is doing this,’ said Francis pointedly.

I was happy with the statement as far as it went, but under its smooth surface cracks were appearing that threatened to blow apart the fragile consensus that Francis had brokered among the HGP participants. What Craig might or might not do or say was no longer the immediate problem. The issue was partly technical and partly political. Our sequencing method depends on having a supply of mapped clones—fragments of DNA from known positions on the chromosomes that are cultivated in bacteria. Pieter de Jong at the

Roswell Park Cancer Institute in Buffalo, New York State, had produced libraries of fragments of about 150,000 bases in BAC clones from the whole human genome, which it supplied to sequencing centers. At the same time an ingenious technique called radiation hybrid mapping, developed by David Cox at Stanford in the United States and Peter Goodfellow in the U.K., made it possible to order markers along the genome very rapidly. These markers could then be used as probes to pick BAC clones for sequencing. David Bentley’s team of mappers used chromosome-specific probes to pick clones belonging to the chromosomes we were working on, and would then ‘fingerprint’ each clone, just as we had done for the worm map. As long as you’ve started with enough clones you can work out their order along the chromosome by comparing overlaps in the fingerprint patterns.

Now you’re in a position to start sequencing. The most efficient way to proceed is to choose a set of BACs that overlap with each other all along the chromosome, but only just—we call this a minimum tiling path. Those BACs are broken down again, this time into pieces about 2,000 bases long, ready to go into the sequencing reactions. At the Sanger Centre we had declared our intention of sequencing chromosomes 1, 6, 9, 10, 13, and 20, and parts of 22 and X, making a third of the genome in total. (As part of the ‘brinkmanship’ that Richard Durbin accused me of, we had made some of these claims even before we knew we had the money to scale up our sequencing—it had the effect of getting others to focus on the necessity of ensuring that all the chromosomes would be efficiently taken care of.) At the various Bermuda meetings others had also staked claims to chromosomes or parts of chromosomes by mutual agreement. Not all of these claims were backed up by funding or capability, but there was an agreed set of guidelines to determine what should be done if someone claimed more than they could deliver. By September 1998, pretty much everything was spoken for. But not everyone who had a claim to a chromosome

was in a position both to map it quickly and to gear up for high-throughput sequencing. And anyone who wanted to get started on sequencing was going to have to wait for someone to produce mapped clones.

Eric Lander was very unhappy about this situation. He was convinced that Craig was indeed capable of delivering the human sequence, and believed that nothing less than a commitment to sequence at the same speed or faster would be an adequate response. The Whitehead’s sequencing center had begun large-scale sequencing work only in 1996, as part of the NIH pilot program, with a project on chromosome 17. More seriously for Eric, it did not have a program for producing a supply of mapped clones that would provide enough raw material to sequence at the rate he now wanted to go. Eric decided that the only solution was to abandon the regional, chromosome-by-chromosome approach to sequencing. Instead, he argued that anyone who had the capacity should be allowed to sequence BAC clones chosen at random from a library of clones covering the whole human genome. He explained his thinking to Francis Collins and Bob Waterston at an informal meeting in Washington in early October. As the first 100 reads from each clone came off the machine, he suggested, you would check them against a central computer database to see if another lab had already covered that clone. If so, you’d chuck it out and start another one. The sequenced clones would be arranged in their correct places on the chromosomes by mapping them afterwards.

Bob was very disturbed by this conversation. It was clear that it was not the first time that Eric and Francis had talked about the plan, and it looked as though other United States labs were also interested. The most disturbing aspect of it was the implied shift in strategy away from the regionally based approach—as in regions of the genome—that had been worked out at the Bermuda meetings, and which seemed to have been accepted at Airlie House just four months earlier. Sequencing clearly mapped regions provides the best

guarantee of the quality of the final product, which is what we were supposedly all trying to achieve. But the approach also made it possible for the project to work as a genuinely collaborative international consortium, with the work shared out among the participants according to their interests and their research capacity. It was open and honest, with everyone knowing what everyone else was doing. Sequencing random B ACs, in contrast, left the door open for those who might want to ‘cherry-pick’ BACs that contained important genes, so that they could get a head start in pursuing commercial developments. And while our regionally based commitments included every step right through to finishing, it was not clear, if everyone scrambled to put out as much sequence as possible from random BACs, who would be responsible for finishing. Eric was apparently unconcerned about this, telling Bob that we could ‘worry about the details later.’

My immediate reaction when I heard about this plan from Bob was that it represented a direct threat to the non-United States members, and also to Bob himself who worked in much the same way that we did. If Eric received the funding for a massive increase in his sequencing output, and if he and other United States labs were able to sequence any BAC that hadn’t already been sequenced elsewhere, they would disrupt our well-organized tiling paths and destroy the efficiency of the process. Our only recourse would be to join in the random free-for-all, which would certainly produce a lot of sequence quickly, but would also present a whole new set of problems for filling the gaps afterwards. And as for the much smaller German, French and Japanese groups, who had enough problems convincing their governments that they should be funded at all, I could not see how they would survive.

Michael had not been copied into the correspondence on this issue over the preceding few weeks, and was furious that such a major change in strategy was being discussed without consulting the Wellcome Trust. NIH and Wellcome had previously agreed that

each funding body would manage the work of its own centers. He wrote to Francis to complain that the agreed strategy was about to be replaced with an ‘uncoordinated, expensive and self defeating free for all.’ Jane Rogers, David Bentley and Richard Durbin were equally alarmed; as the individuals who dealt from day to day with our sequence production and information management, they knew how seriously a major change of strategy could hold up our operation. We made it clear that we saw no reason to change our strategy, but found that our objections were brushed aside as Eric’s proposal gained support. The United States side was by no means unanimous—Bob Waterston was still unhappy about the proposed shift, and Richard Gibbs at the Baylor College of Medicine came down in favor of continuing the regional approach—but it was quite clear that Eric, now backed by Phil Green (with some reservations about the detail) and Maynard Olson in Seattle, had convinced Francis of the need to change course. Their evidence was a theoretical calculation showing that although you might end up sequencing more clones than were necessary with the mapping approach, the costs would be no greater. Towards the end of October most of the United States leaders, some more enthusiastically than others, were beginning to talk about a ‘hybrid strategy’, with some groups working regionally and others randomly. It sounded like a mess, and it wasn’t a compromise I could support.

It seemed to me that Francis was no longer listening to other views. I felt very strongly that to relinquish ‘our’ chromosomes would greatly diminish our influence on the project as a whole. Keeping our chromosomes was about sticking to a scientific strategy that we knew was working in practice, as opposed to one that had so far been developed entirely on paper. We were repeatedly told in the conference calls that both strategies were equally risky. This was simply not true, since we had already sequenced the worm and large tracts of human by mapping first. Crucially, for efficiency the random approach would require the BACs themselves to be dis-

tributed evenly, and this was unlikely to be the case. But it was also about sharing control. That sounds like self-interest, and of course it was, as far as the Sanger Centre was concerned—but, more importantly, I saw the need to protect the international dimension. With a strong international partner the NIH/DOE project could stand up to political pressures to compromise with the commercial sector. Without that partner, it was not at all clear that the whole human genome might not fall into commercial hands. I certainly believed this to be a real danger, and I thought the costs to humankind would be incalculable.

We had a chance to discuss all this at the annual retreat of the Sanger Centre board of management in the middle of October 1998. It was held for the second time at Blakeney, a beautiful place on the north Norfolk coast: a tiny harbor at the edge of a broad estuary that dries to sand and mud at low tide. Huge flocks of birds live there or pass through as migrants, and you can walk out along the low sea wall to get close to them. It’s a wonderful, peaceful spot, where you can look beyond day-to-day detail and see problems in perspective. The directors of the Sanger Centre’s management company Genome Research Ltd joined us: Martin Bobrow, who had moved from Guy’s Hospital to become professor of medical genetics at Cambridge University, and Alan Munro, the Master of Christ’s College, Cambridge. Michael Morgan also came for part of the time.

Rather to my surprise, there was general agreement that the problem might be real: that I was guilty perhaps of mild paranoia, but not without cause. We were conscious of the fact that we were up against powerful economic forces, and that Francis and his advisers had dual loyalties. Under the 1980 Bayh-Dole Act, the National Institutes of Health had a right and a duty to support United States industry by encouraging the commercial development of federally funded research. At Bermuda, meanwhile, they had agreed to collaborate internationally and altruistically. There was no doubt of their desire to collaborate, but there were clearly great pressures on

them; they would surely be looking for a workable compromise, in which the random approach might just be the first step. We in the U.K. had stuck our necks out for internationalism: we were contributing nothing directly to the U.K. economy by our activity (though in the long run our high-profile activity brings indirect benefits). In France, Jean Weissenbach had won a legal battle for French government funds to deliver one chromosome for the HGP. The German government, by contrast, had concluded that its best course was to compete on patentable applications, and phase out its contribution to the Human Genome Project. And indeed our own government had been quite limited in its support. So it was incumbent on us either to ensure that the project remained truly international, or else to admit that the skeptics were right and withdraw from the partnership altogether. As Richard put it, we needed to smoke out what was really going on.

When it became clear that neither Bob Waterston nor Richard Gibbs nor any of us at the Sanger Centre could dissuade Francis from adopting one or other variant of the random approach, I decided that the only answer was to put aside polite scientific discourse and let him know what I really felt. Under the heading ‘Friendly fire’, I e-mailed Francis to express my ‘deep concern’ that he was in danger of accomplishing what Craig Venter and Mike Hunkapiller had failed to achieve: the destruction of the international public program for sequencing the human genome.

I went on to accuse him of orchestrating a U-turn in strategy, and made it clear that I thought his motives were political rather than scientific (meaning that he wanted to strengthen the United States contribution to the project at the expense of his international partners). I asked whether the National Institutes of Health was really interested in pursuing a strong, international program, pointing out that on at least one occasion its press releases had somehow forgotten to mention that the genome project was not an entirely United States operation. I told him that if we could not work round the new strat-

egy we might have to consider going it alone, and implied that as far as the Wellcome Trust was concerned, it should certainly be wary of collaborating with the United States funding agency in future. I concluded with the sentence: ‘I now know something of the feelings of the British tank commander in the Gulf when he watched his crew being destroyed by the weapons of his allies.’

Looked at in the cold light of day, the missive reads as an emotional outburst, which certainly is an accurate reflection of my feelings at the time. But it was not fired off on the spur of the moment. It was drafted over several days, with input from close colleagues at the Sanger Centre (most of whom made me tone it down). I deliberately sent copies to the other genome center leaders and to other interested parties including Jim Watson.

The fallout was immediate. Francis was personally very distressed. He was completely taken aback by the strength of my feelings, and to be honest I had not taken much part in the e-mail debate that had gone on over the previous month, although Jane, Richard and David had made our position perfectly clear on the scientific issues. But he immediately began to look for ways to build bridges, which was exactly what I had hoped for. I was very willing to apologize for the hurt I had caused, but not for sending the letter—I wanted Francis to realize that he could not continue to make policy for the human genome project as if the Sanger Centre did not exist. Several outcomes followed in quick succession: a somewhat frosty conference call with the rest of the genome center leaders, and then an invitation to meet the head of the NIH, Harold Varmus, in London, and to go to Washington a month afterwards to address the funding agency’s Scientific Advisory Council. The letter was clearly being taken seriously.

The London meeting, hastily fitted in to Varmus’s already busy schedule, took place over breakfast in a café in Soho, horribly early on a dismal November morning. Martin Bobrow accompanied me from Cambridge and Mike Dexter, who had recently taken over

from Bridget Ogilvie as director of the Wellcome Trust, also joined us. I argued that the Sanger Centre and Bob Waterston’s lab in St. Louis were ready and able to produce mapped clones for sequencing, that currently only our two labs and Celera had put in place the high level of industrial organization needed to accelerate the production of sequence, that it would be unwise to try to implement a major change of strategy under the sort of time pressure Celera’s appearance had imposed, and that the Sanger Centre was committed to finishing its chromosomes. I didn’t feel at the time that we made much headway. Varmus seemed unwilling to make a complementary commitment on behalf of the United States labs to finish the chromosomes not spoken for by non-United States labs (although he later explained that he could not do this without consulting Francis). He queried the fact that I had not included the Whitehead Institute in my list of centers that were ready for a massive scale-up, which confirmed what I had begun to suspect: that despite the relatively modest contribution (compared with St. Louis and the Sanger Centre) he had made to date, Eric Lander had consolidated his position in the eyes of the NIH hierarchy as a leading figure, perhaps the leading figure, in human sequencing.

I wasn’t sure about going to the Advisory Council meeting—I felt that I was being summoned to give an account of myself. It was Jim Watson, who had been very active behind the scenes in opposing the random clone strategy and retrospectively approved of the ‘friendly fire’ e-mail, who sowed the doubt in my mind: ‘Don’t go to Washington!’ was his advice. But as the date in early December drew nearer, it seemed there was more to be gained by going than not. The message that high-level international consultation was important seemed finally to be getting through, and a meeting of the heads of the five biggest sequencing operations—Richard Gibbs from Baylor College of Medicine in Houston, Bob Waterston from Washington University in St. Louis, Elbert Branscomb from the Department of Energy’s Joint Genome Institute in Walnut Creek,

California, Eric Lander from the Whitehead Institute in Cambridge, Massachusetts, and me from the Sanger Centre—was scheduled for the same visit. The fact that Bob Waterston would be there reassured me that I would not lack support. Michael Morgan was also invited to talk to Francis Collins.

The meeting of the five labs started a tradition of regular communication among what was first jokingly called the ‘security council’ of the Human Genome Project, and later became the G5. I was determined to do whatever was necessary to keep the international collaboration going, without compromising the Sanger Centre’s position on the allocation of chromosomes. Bob and I offered to solve the problem of the supply of mapped clones by acting as a resource for the whole community, providing clones to other labs for sequencing. Eric accepted the offer, with the proviso that if the supply of mapped clones could not keep up with the pace of sequencing he should be free to choose BACs at random. (In the event, Bob’s lab took over the clone supply for the United States, and eventually, under Marco Marra and John McPherson, established a central map to coordinate the entire project, which saved it from chaos. I was unable to persuade the Sanger Centre mappers to give Eric many clones at all, which was embarrassing but not too surprising, especially as the Wellcome Trust had always discouraged us from mapping for others.) The American funding agencies finally committed themselves to ensuring that the rest of the chromosomes were finished. All in all, it was a highly productive meeting.

The next day’s presentation to the Scientific Advisory Council of the NIH was a formality by comparison, though politically important. I felt uncharacteristically nervous in the big room with its central table and the staffers arrayed in ranks around it. I had been summoned and had obeyed, just as a century earlier colonial leaders had been summoned to London and had obeyed. There were no tapestries or banners, indeed no decoration at all in the utilitarian

government-issue room, but this was a theatre of power in the greatest empire on earth today. The NIH had an annual budget of $15 billion, twenty times that of the Wellcome Trust: why should they listen to me?

The Council was meeting to ratify Francis’s plans for distributing the funds awarded by Congress for the genome project. Francis had presented the strategy for completing a draft of the genome by 2001 and finishing by 2003, as announced in October. My role was to stand up and say that Britain, in the form of the Sanger Centre, would support this strategy. Simple enough to do; but a lot was at stake. This was also the first occasion since the Celera announcement on which I had met Craig Venter face to face—in fact, I don’t think I’d seen him since the 1996 Bermuda meeting. It wasn’t that I had particularly been avoiding him, just that we had had no reason to meet, and I had certainly seen no point in trying to establish any kind of dialogue.

There had been a phone call soon after the Cold Spring Harbor meeting, though, a conference call out of the blue from Craig and Mike Hunkapiller to sound me out about their announcement. ‘This isn’t an in-your-face thing,’ they said: they really wanted to get the sequence out for everyone. Indeed, would I be interested in joining them? I hesitated for a moment; nothing is absolute, after all, one must consider every case on its individual merits. How tempted Francis and Harold must have been—to avoid the conflict, make the compromise, get the job done without a fuss and live peacefully. And then I remembered the track record: the attempt of ABI to control assembly software, the human ESTs for sale, the forcing out of Jim. Wherever these two people had been, the decisions came down to profit. I was back in the Claremont tower. ‘No, thanks,’ I said. We talked for a little longer about their plan to build the best possible database, and a thought occurred to me, not for the first time. ‘You’re trying to be a Microsoft of biology, aren’t you?’ I said. ‘Oh no,’ was the reply, ‘it’s not like that at all.’ A little while later, I started to see articles comparing Craig Venter with Bill Gates. Some

might see this as a good thing, but the genome is a unique resource, not a technology.

Craig sat down next to me at the Advisory Council meeting; I greeted him and shook his hand, but received no sign of friendliness in return. It was very strange to be sitting next to the man I had come to regard as the potential destroyer of all that we had worked for.

Craig was there partly because of another development, even more serious than the random clone business, that threatened the fragile alliance among the publicly funded sequencing labs. Unknown to Francis or any of the groups funded through his institute, Ari Patrinos at the Department of Energy had drafted a memorandum of understanding with Celera to provide Craig with BAC clones in return for collaboration on the sequencing of the Joint Genome Institute’s human chromosomes 5, 16, and 19 and equivalent regions in the mouse, with the specific aim of looking for gene families that might make important drug targets. Francis had got to hear about this, and had let the rest of us know, only a couple of weeks previously. Clearly the Department of Energy was under political pressure to collaborate with the private sector; but the terms of the draft memorandum with Celera contravened the Bermuda agreement on free release of data to which the department’s scientists had signed up.

Under pressure from the rest of the group leaders at the ‘security council’ meeting the day before, led by Eric (who gave a magnificent and passionate denunciation), the group funded by the Department of Energy withdrew from negotiations with Celera. So it was not surprising that Craig did not look best pleased when he appeared at the Advisory Council meeting, particularly as he had been told he could not speak. At the last minute he was invited to comment on reports of a change in Celera’s business plan. He complained that he had nothing prepared, but said that nothing had changed since his previous testimony, and that if there were to be a change it would be announced. He declared once more, as he had at the congressional

subcommittee in June, that he had no intention of knocking the public effort, and earnestly wished to work alongside it. But he seemed unable, or unwilling, to appreciate that his commercial ambitions would be in conflict with the HGP policy on data release and therefore an insuperable stumbling block.

I returned to the U.K. feeling better about the whole thing than I had for a long time. Maybe the DOE—Celera episode had made everyone realize the absolute necessity of the publicly funded teams working together. To have seen off that threat, potentially found a way round supplying mapped clones to sequencing centers and presented a show of unity to the great and good of the NIH was, all in all, a good couple of days’ work for a project that had looked in danger of fragmentation on more than one occasion in the previous months.

And I had a celebration to look forward to. Nine and a half years after the moment when the prison door clanged shut at Syosset, the worm sequencing consortium was to publish the sequence: the first of an animal, and indeed the first of any multi-celled organism. It came out in Science during the second week in December, and the NIH and MRC held press conferences on each side of the Atlantic to mark the occasion. Bob Horvitz remembers the event vividly—he was at the Washington press conference.

In his talk, Bob Horvitz said the achievement was ‘better than landing on the moon’: both were human achievements, but the worm genome sequence was an achievement you could use for science and medicine. On our side we went down to the Royal Society with a group of people from the lab. I talked about the sequence and Jonathan Hodgkin, my contemporary from the early days with

Sydney who still worked on worms at the LMB, talked about its implications for worm biology.

The truth is that getting the worm done had been an underlying objective for me all the way through. Of course I had my day job to do, directing the Sanger Centre, and I was always there when the HGP needed me; but all through the alarms and excursions of 1998—the doubling of our goals for human sequencing, Celera, friendly fire—I was spending time in the lab helping to close gaps in the worm sequence. And between us Bob Waterston and I and everyone else involved made sure that the worm got done. That was our own: if you don’t do your own project you have nothing. If you can make something else happen as well, then that’s great, so I don’t regret becoming involved with the human genome. But I definitely had a sense, at the end of that year, of having achieved what I had set out to achieve. Even then, the worm sequence still had some tricky gaps in it, and I looked forward to working on some of the remaining difficulties during the coming year.

The worm had been successful in a larger way, too: it had been one of the sources, right from the Room 6024 days in the early 1980s, of a new field of biology. That little trickle had become a stream, and had joined other streams to form the river that is called genomics and is transforming the way people do biology—not by displacing real experimental work, but by empowering it with new tools and entry points, with access to the totality of the codes of life. It was all working, and so for me, just like the time at the end of the cell lineage, it would soon be over.

This was different from the last time, though. The scale was vastly larger, for one thing. But more importantly, and partly driven by genomics, biology had undergone an economic sea change—it now held the promise not only of tremendous knowledge and great benefits to humankind but also of fabulous wealth. As biologists we had lost our innocence; we were out in the so-called real world and were reacting to it in a variety of ways. Before, I hadn’t had to face

up to the true extent of venality in the world, because it hadn’t impinged on me directly. Now it began to look as though I wouldn’t be able to do science any more, but would have to join some ill-defined political struggle, to try to make some contribution, however small, to turn the tide.

In practical terms I had a sense of having discharged my duty to the lab and the Trust by having at last secured the funding for a third of the human genome. This was the guarantee that the genome would remain international and that the Trust would come out with honor, having secured its own primary objective in funding the Sanger Centre. Jane was firmly in charge of sequencing (as well as being overall project manager for our part of the Human Genome Project) and it was going better and better all the time: I no longer went to many meetings, feeling more useful out of the way in the lab. And so it seemed the right moment to give in my notice as director.

It wasn’t a spur-of-the-moment decision. I’d been quite open with Richard, Jane, David and my other close colleagues about wanting to step aside as soon as things were running smoothly. At the end of August I had written to the BoM as well as the directors of Genome Research Ltd, telling them in confidence that I wished to make an announcement by the end of the year, and aimed to be replaced by the end of 1999. These things take time, and Alan Munro warned me that two years would probably be more realistic, so it was time to start the process. I reminded them that I had always been a reluctant director, and that my particular interests meant that I was not going to be part of the post-sequencing phase of biology. ‘In the changing environment of biology today,’ I added, ‘my instinctive reactions are becoming less and less appropriate.’ What I meant was that I foresaw a time when centers like ours would move on from producing genome sequences for the benefit of all to a more targeted approach to understanding genes and their products. It seemed inevitable that such developments would involve questions of intellectual property

and commercial development which had never formed part of my thinking.

I was confident that with the distributed system of management we had in place, the Sanger Centre was in robust shape and would not be in any way damaged by my going. Indeed, a change of head would be helpful; the genomics field was moving so fast that the lab needed to be renewing itself by 2000 if it was to retain credibility. But equally it was obvious from the reactions to my confidential letter that unless I forced the issue the Wellcome Trust would keep jollying me along and not do anything about finding a successor. They issued their announcement in the middle of December. I had been right to get things moving early—my successor did not take office until almost two years later.

One thing was clear: I had no intention whatever of becoming a public face of the human genome.