I WAS SHAMELESSLY UPSTAGING THE REST OF THE GROUP. FROM MY wheelchair I was trying to persuade the director of the Wellcome Trust, Bridget Ogilvie, as well as Michael Morgan and other Trust representatives, that we should accelerate our sequencing program immediately and cover most of the human genome by 2001, four years ahead of schedule. It was 1 December 1994. Five days previously I had been knocked off my motorbike and carted off to hospital unconscious, with a broken pelvis. Yet I was so carried away with excitement over this new idea that nothing was going to stop me—and it seemed that the Trust group was excited too.

It was Bob Waterston who had conceived the bold plan to accelerate the project. At the beginning of September 1994 he came over with his colleagues from Washington University for the annual joint lab meeting—‘that fateful visit’, he calls it. Bob found me rather worn out by politics and doubtful about how to go forward. As always, he stayed with Daphne and me, and we talked a lot.

It had been a confused year. The Sanger Centre was growing rapidly, and was becoming visible to the world of science. It was

increasingly evident that we were not going to be a flash in the pan, and of course this meant that rivalries began to emerge, as in the case of chromosome 22. It was clear that if so much fuss could arise over just 1 percent of the genome, we were not going to get the job done if we carried on in this fashion. These were early days, though, and we still had a little time to sort out the politics. So it was well worthwhile continuing to try for general agreement.

Important as it was to get the human sequencing onto a workable footing, my own first priority was to get on with the worm sequence. Worm and yeast were the Sanger Centre’s lead projects in terms of both our scientific credibility and technological development. But the worm work was funded wholly by the MRC. The Wellcome Trust, which at that time was putting up an imposing new building to be the Sanger Centre’s permanent home, wanted to see more human sequence as soon as possible. Only human sequence would justify the large investment in such a high-profile project as the Wellcome Trust Genome Campus. However, it was too early: mapping was going forward, but space in the old building was limited, and there were as yet too many problems with finishing human sequence for that part of the operation to be easily accelerated. We were actually following quite closely the schedule in our original proposal, but there was pressure to do more. I’d given Bob some idea of my anxieties when we talked on the phone before he came over, but this was the first chance I’d had to discuss it face to face. He sat for a long time at the picnic table in my garden one late summer evening, listening sympathetically and obviously concerned that I seemed to be under so much stress.

Soon after Bob returned home to St. Louis, he sent me an e-mail headed ‘an indecent proposal.’ It contained nothing less than a strategy for completing the human genome in what by then current standards seemed an impossibly short timescale. It set out over several pages how we could each accelerate our production of sequence to 200 megabases per year (at the time we were doing ten or fewer). It went into great detail about the number of machines

and people we would need, how often we would run the machines, and what it would all cost. It departed from our previous practice in proposing that we should churn out sequence as fast as possible and use automatic assembly and editing procedures to string it together to a reasonable but not absolute standard of accuracy—99.9 percent rather than 99.99. We would continue the slower process of hand finishing, but initially at little more than our current rate of about 10 megabases per year. In other words, he was talking about beginning with a working draft of the genome (but one much better than the one the world celebrated in 2000), and moving on to the gold standard, finished product only later. Bob reckoned we could get 200 megabases of almost finished sequence and 10 of finished sequence for a budget of $20 million per year, an average cost of around 10 cents per base—an order of magnitude less than the costs that were being discussed a decade before. Bob concluded, ‘I would propose that you do 200 [megabases], we do 200 [per year] and we think about involving a third group and propose to do the entire genome between us in 5 years starting in 1996.’

It turned out that the idea had taken root in Bob’s mind on the long, boring transatlantic plane flight back to St. Louis. He had a watch that had calculator buttons.

Rick Wilson bought it, though he commented sardonically that the St. Louis lab would stop Bob taking his calculator watch on airplanes in future: the combination of the buttons and an oxygen-starved Bob brain was clearly dangerous for everyone’s sanity.

I too realized immediately that we could do it. In the Genome Sequencing Center at Washington University in St. Louis and the Sanger Centre at Hinxton, Bob and I headed the two most productive genome sequencing laboratories in the world. We each presided over an operation involving a couple of hundred people and dozens of machines, working a factory-like schedule to keep the sequence pouring out seven days a week, but at the same time constantly refining the tools needed to analyze it and improving the technology that generated it. By 1994 we were speeding up: at the Sanger Centre we were going to meet our target of 40 megabases of finished human sequence in the first five years, as well as finishing the worm and yeast genomes. Automation was improving all the time.

A crucial advance had come from St. Louis. Since 1989 Bob had been collaborating with a colleague in the Genetics Department, Phil Green. Phil originally trained in mathematics and was interested in taking a mathematical approach to solving problems in biology. When we started sequencing the worm, he immediately began to develop programs to help with assembly and gene finding. By the end of 1994, when we were thinking about scaling up our work on human DNA, he had designed a piece of software known as phrap (Phil’s Rapid Assembly Program) that could assemble human sequence far better than previous assemblers. Assembly software for the human genome has to take note of the fact that half of it consists of bits of DNA that look similar, known as repeat sequences. Because of these repeats it is very easy to make errors in assembling the raw reads into a continuous sequence, but Phil’s program makes sure that you get the best possible match for each repeat. In 1994 Phil had just left Washington University and gone to the University of

Washington in Seattle to join Maynard Olson; there the two of them lent their highly influential and independent voices to the debates over the strategy for the human genome that took place in succeeding years.

It wasn’t just that we could get going on the human sequence; Bob and I both felt strongly that it needed to be done. Although the HGP had committed itself in 1990 to sequencing the whole human genome by 2005, it had begun with a raft of projects, including the worm sequencing effort, that did not produce human sequence at all. Less than 1 percent of the genome was represented in the public databases by the end of 1994, and most of that was in the form of cDNAs rather than genomic sequence.

Meanwhile the EST approach to finding genes was making rapid strides. For two years Craig Venter at TIGR and the Silicon Valley-based company Incyte Genomics had each been building up databases of these partial gene sequences. TIGR’s arrangement with its associated company, Human Genome Sciences, meant that although academics could look at the ESTs in the TIGR database, HGS—or rather, its exclusive licensee SmithKline Beecham—had six months to a year of exclusive access to them first. Incyte’s database was a proprietary product that you had to pay to use, and they also retained rights on further development. Although the NIH’s ill-advised bid to patent the ESTs that Craig had discovered had been withdrawn, it looked as though private interests were finding other ways of locking up basic information about the human genome, moves which could only have a detrimental effect on future discovery.

In 1994 a white knight had come along in the form of the pharmaceutical company Merck, and in particular its vice-president Alan Williamson. The big companies weren’t any happier than the academics that upstart genomics companies looked like cornering all the rights to valuable genome information. Merck funded a massive drive to generate ESTs and place them in the public databases,

where they would be freely available to all. The effort involved a consortium of United States public labs, with the Genome Sequencing Center at St. Louis at its heart. Bob Waterston and Rick Wilson received a grant from Merck to generate 4,000 ESTs a week for two years, starting in January 1995. By doing this, Merck not only gave the entire research community, public and private, free access to valuable genomic data; it also made those sequences (and possibly the whole genes from which they came) much more difficult to patent. Once the sequences had been in the public domain for a year they could not be patented; and it would be tough for any company to identify the most promising genes out of so many and understand their function in such a short space of time.

I thought Merck’s action was a great thing for science and a triumph for the principle of free access to genomic information. But Bob and I both knew that the EST database would not be enough to reveal all the genes, never mind to understand how the genome functions as a whole. The strategy depends on finding enough RNA to make the complementary DNA strand. Some genes are expressed in such small amounts, or such specialized tissues, that it is hard to find the RNA at all. But, much more important, RNAs correspond to only about 1.5 percent of the total DNA. The rest may not be directly involved in coding for proteins, but it is certainly not all junk. In order to understand how living tissues make use of genomic information in an integrated way, you have to do the whole thing. It had been hard enough to win the argument for whole genome sequencing back in the 1980s. Now, with the databases filling up with ESTs, there was a danger that people would question once again the value of sequencing the ‘junk’ between the genes. The best answer would be to do it, and to show how fantastically useful it was to scientists trying to understand the basis of disease.

Why wait? The HGP funders were hoping for some magic new technology to come along, and putting a lot of money into projects that promised to develop it. But meanwhile we were finding that the

existing technology could do the job perfectly well. Running gels might seem fiddly and labor-intensive, but we were constantly finding better ways to use them. ABI was promising machines with more and more lanes—already we had modified their machine to run forty-eight lanes, and sixty or even ninety-six lanes were talked of. They and others were already beginning to experiment with capillary technology—running each sample through a fine, gel-filled tube instead of a lane in a slab of gel. There were people working on automating all the tedious and time-consuming jobs such as picking clones. We had excellent support from our bioinformatics teams in developing innovative software to track the samples and make sense of the results. It was all working.

Even so, I gulped a bit when I saw what Bob had written. But I quickly realized that it was not ridiculous. Bob is not a man given to wild impulses; on the contrary, he’s a wonderfully stabilizing influence. His confidence was infectious. It immediately pulled me out of my gloom and set me thinking about how we could begin to set in train what I dubbed the MGP—the Megalomaniac Genome Project.

We knew from the start that one obstacle would be political. Although we were the biggest producers of sequence around, we were known as worm people. There were labs all over the world that had been mapping human chromosomes, more or less effectively, for years. They would undoubtedly expect that in the fullness of time they might move on to sequencing. Many already were, on a gene-by-gene basis, and one or two had funds to begin large-scale genomic sequencing. We could well be seen as interlopers, and resented accordingly. I saw this as being a particular problem in relation to the other international genome labs, such as those in France, Germany and Japan, and was anxious that any formal proposal should not be seen as too exclusive. Bob took the point, but put the alternative view that it would be impossible to accommodate the interests of everyone who thought they deserved

to be involved. He pointed out that our bipartite collaboration on the worm was exceptional among genome sequencing projects in its success and lack of friction. And he admitted that the megalomaniac aspect of it was part of its attraction; sounding out colleagues such as Rick Wilson, he’d already found that the very audacity of the plan could be a powerful motivating force. But he agreed that trying to do the whole thing between the two centers would upset a lot of people, so we stuck with the proposal that we should do a third each, leaving a third for others.

We also knew that we had a practical problem. In the case of the worm we had a collection of mapped clones covering the entire genome, all ready to go into the sequencing pipeline. For the human, such a comprehensive set of suitable clones did not exist. There were technical problems in getting the DNA to clone stably in bacterial cosmids, as we had done for the worm. Although there were YAC maps in existence, they were of variable quality. The first whole-genome physical map of the human genome had been created by Daniel Cohen at Généthon in France. Cohen, who was director of the Centre d’étude du polymorphism humaine (CEPH) in Paris, launched Généthon in 1990 with funding from the French muscular dystrophy association as one of the first large-scale, automated genome ‘factories’ in the world. (The launch of Généthon had been one of the factors that predisposed me to the idea of a development such as the Sanger Centre in the U.K.) Cohen had moved fast to produce YAC maps, first of chromosome 21 and a year later of the whole genome. At the same time, his colleague Jean Weissenbach had produced a whole-genome genetic map, three years ahead of the timetable set by the HGP. But while the genetic map has remained one of the main anchors of the project all the way through, the physical map turned out not to be such a useful resource.

The problem was directly related to the nature of human DNA. We had been able to use YACs for the worm map, once we had sorted out problems of contamination with yeast DNA and learned

to deal with the large size of the inserts. But the number of repeats in human DNA—as much as 50 percent of the total is in repeat sequences, compared with 15 percent in the worm—caused constant problems in getting stable clones. They would break up and rearrange themselves as the yeast cells reproduced. The Généthon map had particular problems because Cohen had developed a method of making ‘megaYACs’ up to 1 million bases long, which only increased the opportunities for chimaeras and deletions. New bacterial cloning methods had recently been developed, such as bacterial artificial chromosomes or BACs, which held smaller inserts and looked as if they would clone more stably. But there was a lot of work to do to produce a library of BACs and map them before they were ready to go into the sequencing machines. It was something we thought we’d be able to crack if we put our minds to it, although we were admittedly a bit vague on exactly how to go about it. We reasoned that there was no need to complete the map first, as we had for the worm; we could map and sequence in parallel.

Obviously, we could not launch such a huge effort on the basis of the funding we had. I was fairly optimistic that we at the Sanger Centre would be able to get what we needed, as the Wellcome Trust had gone into genome sequencing with the clear intention of doing substantial amounts of human sequence. I hoped it would just be a question of persuading them to bring some of the money forward. For Bob it would be more difficult. He had already had to fight a battle with the NIH, which had upped his funding for the worm in 1993, to be allowed to spend some of his grant on small amounts of human sequence, in parallel with what we were doing—the study section that reviewed his proposal did not believe that the methods would work on human DNA. ‘In the end they more or less threw up their hands and said they didn’t think we should do human, but it would be up to us to decide,’ says Bob. ‘And so we did a little bit of human—but we knew we were being watched.’

Because of the political sensitivities we didn’t want to make a big splash with our idea until we’d done a bit of informal sounding-out. Among others Bob talked to Francis Collins, who had by then been head of the NIH genome program for a year, and Maynard Olson, who could be relied on to spot any flaws in the plan. ‘Nobody could shoot it down,’ he says, ‘although Francis was very cautious.’ They all seized on the problem of clone supply, which of course was our weakest point. On the political question, there was a widespread view that no one else was yet in a position to take on the remaining third of the task. Meanwhile I talked to Jane, Richard and David, since they would have to bear the brunt of managing a ramp-up in sequencing, data management and mapping. They were immediately enthusiastic, and in a very short time Bob and I had arranged to make presentations to the Wellcome Trust and the Medical Research Council in the U.K., and to the National Institutes of Health in the United States. Bob was going to come over and back me up at the Wellcome and MRC meetings.

On the Saturday before the date set for the meetings in London, I set off from home for the lab on my motorcycle at about seven-thirty in the morning. I was still a member of one of the worm sequencing groups, and took my turn loading machines at weekends. I had ridden a motorbike as a research student, and had recently taken it up again; the LMB was only a short cycle ride away from Stapleford, but Hinxton was further from home, and I also liked the fact that I could get to the airport on the motorbike more quickly than I could by car. That morning, at a junction a mile into my journey, a van ran into me from the side. I remember nothing of the accident, but evidently the impact pushed my leg back through the pelvis, breaking both the pelvis and a ligament in my knee. My 550cc bike was left as little more than nuts and bolts, and I was flung right across the road. But I was incredibly lucky. A woman in the car behind phoned for an ambulance and I was taken straight into the operating theatre at Addenbrooke’s Hospital in Cambridge. A&E was empty at that

time in the morning, and the duty surgeon was Dennis Edwards, one of the best in the business. The nerves were intact, the injured tissue had had no time to die, and he reset the pelvis in perfect alignment, holding it all together with a plate and screws, which I still have.

When I came round, still groggy from the anaesthetic, I tried to get out of bed and go and load the sequencing machines. Then when I realized I couldn’t move, I thought I had to try and find someone else to go and do it. That may have been a little irrational, given what had just happened. But as my mind cleared, I was gripped by the thought that I had to get to the meeting at the Trust on Thursday. As soon as I was conscious Bob came on the phone, trying to get me to give some sort of account of how badly hurt I was. I just kept saying, ‘Oh, it’s OK, this can’t stop us, there’s no way we can change our plans now—when are you coming over?’ He agreed that he would come, although he says now his main anxiety was to find out how I really was. I spent the next three days battling with the doctors to let me leave hospital in time for the meeting. In fact I made such a nuisance of myself in the ward, constantly using the phone, holding meetings with colleagues and so on, that they put me in a private room. They said I couldn’t leave until I could get around on crutches, so I practiced doggedly until they reluctantly agreed to let me go. On the morning of the meeting I got myself up in the pitch dark, dressed and maneuvered myself into my wheelchair with my crutches, and went down in the lift. Bob was staying with Daphne, and David Bentley was there overnight too, ready to go down to the meeting the next day. They had arranged a car and a driver to take us to London, and the hospital discharged me into Bob’s care. I told them he was a doctor, which was true, although he hadn’t practiced since finishing his training.

And so I arrived on time at the Wellcome Trust meeting—thanks to Bob, David and the wheelchair. I was so hyped up I felt I could have sold anything, and the meeting seemed to go very well—well

enough, anyway, that they invited us to submit a formal proposal. We left in high spirits, Bob pushing me down the Euston Road to the MRC, stopping for a pub lunch and a quick turn round the rose garden in Regent’s Park on the way. Bob thought the afternoon’s meeting was less successful than the morning’s. But the MRC secretary Dai Rees gave us a fair hearing, and I certainly felt that we were being taken seriously. All in all it seemed like a good day’s work.

The surgeon, Dennis Edwards, had promised to have me back on the bike within six months; I was more concerned about being able to go hill walking. Ever since I was a student I’ve got out for a few days’ walking each year, often in the Lake District or Scotland, and lately my son Adrian and I had taken to going together for a long weekend in the spring, scrambling around the crags of the west coast. But I need not have worried. Just the following summer Daphne and I stood on Mount Brandon, 3,000 feet up, looking out over the Atlantic in south-west Ireland; I sensed that Edwards’ handiwork was perfect, and that he’d given me back the beauty and solitude of the high places. Adrian now lives in Edinburgh, where he works in software, and from there we continue to have wonderful excursions.

I didn’t replace the bike. Bridget Ogilvie and her colleagues at the Wellcome Trust were horrified when they discovered I’d been using it to get to the airport—‘After we’d invested all that money in this bloke!’ she says—and forbade me to do it any more. More significant from my point of view was that Daphne, too, felt that this had gone far enough. But I didn’t need much persuading. It makes life pretty important that you’ve survived something like that. I don’t mean that I had any religious sense of having been preserved for a purpose. But I certainly felt that I was very lucky—not only to be alive, but to have been put back together again as good as new—and that I had better make the most of it. I was back at the lab within a week, and turned in my crutches three months later. In the interim, I shared a

taxi into work with Matt Jones, one of the group leaders at the Sanger Centre, who had been knocked over not long after me and was recovering from a fractured femur.

Two weeks after our London meetings, Bob presented our idea to a gathering of genome lab heads at Reston in Virginia, which had already been scheduled to discuss future strategy on the human genome. Most of them had no inkling of what we had planned; we’d taken only a few into our confidence. Bob was down to speak, and naturally they assumed he was going to talk about progress on the worm genome. Instead he produced a set of handwritten overheads he’d made the night before and laid out our proposal. ‘They were taken aback,’ says Bob. But Maynard Olson stood up and said that what was on the table was a realistic plan to sequence the human genome, the first anyone had produced, and that they should take it seriously. The debate that followed focused on two issues: the map, which Bob knew was a problem (‘I didn’t quite say so, but I thought that was their responsibility!’ he says) and the value of having a draft—an intermediate product on the way to fully finished sequence. Yes, the worm community had confirmed that unfinished sequence was useful. But there was a real worry that once the draft was available everyone would lose interest in finishing, and we would never end up with an archival product that would stand the test of time. On the other hand, Francis Collins had seen that the plan provided a lever with which he could increase his budget, and he had arranged for Bob to talk to Harold Varmus, the NIH director, just before the Reston meeting. Francis knew that sooner or later there would have to be more money for production of human sequence. The wide distribution of smallish grants for mapping, technology development and small-scale sequencing projects that he oversaw at the time would not meet the need. Even giving everything in the pot to a smaller number of people would not solve the problem; the pot itself needed to be bigger.

Now that the plan was out in the open, everyone was talking about it. The important point was that no one seriously challenged Bob’s math. His timescale for getting a 99.9 percent accurate sequence seemed credible. But in addition to the mapping and finishing anxieties, political worries began to circulate, and these were even worse than we had expected. If the funding agencies put so much money into a few big centers—not necessarily the same ones that were doing the mapping—what would happen to the rest of the twenty or so labs in the United States (and a similar number in the rest of the world) that got left out in the cold? In an interview about our plan published in Science magazine in February 1995, I argued that, given that money was tight, we had to go for the most cost-effective option. ‘The entire bill for biomedical research would be lower in ten years’ time if we start now than if we delay,’ I said. ‘If we don’t start now, there will be innumerable other gene hunts and sequencing projects going forward, taking collectively an enormous amount of money.’ If it can be done now, I argued, ‘Why fiddle around?’ Our thinking at the time was that as the draft was produced, anyone who had an interest in a particular region could get the relevant clones from us and finish them to a higher standard.

The debate continued at the May 1995 Cold Spring Harbor genome meeting. The organizers (who included Bob Waterston and David Bentley) decided to break with tradition by having not one but two ‘keynote speakers’ to round off the meeting just before the farewell banquet on the Saturday evening. They invited Maynard Olson and me, Maynard to speak about mapping and me about sequencing. Maynard spoke first. I had been a bit nervous about what he would say, because despite his support for Bob at Reston, he had certainly been very doubtful about the idea at the beginning of the year. Always a perfectionist, he had been afraid that producing the draft would be a distraction from the real goal of fully finished sequence; and he thought that if money were taken away from developing new technologies to put it into production

sequencing we would never reap the benefits of a fully automated process. But by May he’d had a chance to talk it over with Bob, and was ‘generally satisfied’ that our plan would not undermine the long-term goal. Maynard and I shared a house on the Cold Spring Harbor campus during the meeting, and had a chance to coordinate our messages. We sat at the same table preparing our talks, with me scribbling out my usual handwritten overheads, he going off to the conference office to get his properly printed.

Maynard backed us in his talk, though still expressing reservations about the quality of the product. I was then able to get up, lay out our plan to produce what we were calling a ‘sequence map’ by 2002, and say: ‘Let’s just do it.’ (To Maynard’s mortification, his beautifully produced slides were printed right to the edge of the sheets and did not fit on the screen. I could not resist using the incident to contrast our styles of work. ‘My slides may not be as smart as Maynard’s,’ I said, ‘but at least they fit on the screen!’)

Not everyone was convinced. For instance, Eric Lander, director of the Center for Genome Research at the Whitehead Institute, was standing next to me in the lunch queue one day at the meeting, and told me that he thought it was too soon to launch a bid for the whole genome. Eric is a big, ebullient New Yorker with ambitions to match. He had started out as a mathematician and taught for a while at Harvard Business School, but in the 1980s he had switched to medical genetics and joined the Whitehead Institute. He became convinced that the quickest way for the field to move forward would be to make a big investment in infrastructure, in the form of genetic and physical maps. He worked with Helen Donis Keller and Phil Green, who were then at the private company Collaborative Research Inc., in the first attempt at a genetic map of the human genome, placing about 400 markers in total. In 1990 he received funding in the first round of HGP grants to establish a Genome Center at the Whitehead and to make first mouse and later more human genomic maps. The center had established a high profile

with an approach that depended on large-scale automation. In 1995 he began to turn his attention to high-throughput sequencing, the obvious next step.

At the time of our proposal, Eric had only just begun to plan for sequencing and would need more time to develop the same kind of highly automated approach to sequencing that he had taken to mapping. At the same time he was critical of our proposal because we wanted to continue on our course, automating as we went along. Through our success so far we had demonstrated that this approach worked. Each year our production had gone up and our costs had gone down, and naively I assumed that our results would speak for themselves. But Eric’s doubts would carry weight—much more weight than I realized at the time—and he was not alone in his views. Today he still argues that our bid was premature.

Eric may have needed another two years, but we didn’t; we were ready to go ahead as we were. And, despite his reservations, Eric began to talk to us about joining our collaboration—in effect, becoming the third partner that our proposal envisaged. In July we had a formal exchange of letters, agreeing to share expertise and to hold small group meetings of people from the three labs who were working on the same problems.

Eric is the first to admit that ambition is one of his defining characteristics.

Meanwhile we knew what we could do, and didn’t think Eric’s criticisms were justified. We pressed on with preparing our proposal. Jane, Richard, David and the rest of the BoM put together a massive document outlining a program of work for the Sanger Centre from 1996 to 2002, including sequencing one-third of the human genome. Of course, this included our own plans for automation which would go on alongside the scale-up. I added an introduction.

I sent the proposal, asking for an additional £147.2 million over seven years, to both the MRC and the Wellcome Trust. This was for fully finished sequence. Everyone agreed that the MGP was only a step on the way, so it made sense to budget for the finished product. Asked by the MRC what would be the consequences if we were funded to do one-sixth of the genome, only half as much as we had projected, I replied, ‘In such a climate, the Sanger Centre’s influence in driving forward the international program would be lost. As a result of this it would be unlikely that the major uncharted regions of the human genome would be completed systematically and in the public domain in the foreseeable future.’

The proposal came up for consideration by the Molecular and Cellular Medicine Board of the MRC, and the Scientific Committee of the Governors of the Wellcome Trust in the autumn of 1995, a year after Bob and I had first begun to put our plan together. At the end of September a joint MRC/Wellcome committee descended on the Sanger Centre to review our work and collect information for a report to the two funding bodies. Mike Dexter, who was then deputy director of the Paterson Institute of Cancer Research in Manchester and chairman of the MRC’s Molecular and Cellular Medicine Board, chaired the delegation. It included, as an ‘observer’ for the NIH, Eric Lander. I was somewhat surprised when I heard he was to take this role, given that he had already signed an agreement to collaborate with us.

It was a strange occasion. We had invited Bob to come over, to emphasize our international collaboration, just as he had invited me for his NIH review board a few weeks previously. There I had sat at the back most of the time, but I had been called to get up and say a few words about how much we valued the link and what we had been able to achieve. But at our review Bob was not allowed to

speak. The committee sat behind a long table in the rather nice conference room that we had inherited from Tube Investments at Hinxton. At that time there was a vogue among accountants for large chunky calculators with big square keys. I always thought on seeing these things that the message was: ‘I have big square buttons and I deliver big square numbers that are better than your little round numbers.’ Murray had one, in keeping with his position, and so did Eric on this occasion. He used it repeatedly to challenge our calculations, casting doubt on our ability to scale up the production of sequence on the timescale we envisaged and questioning our cost estimates.

I was confident that the figures Jane had prepared for the proposal, based on her considerable experience of running a sequencing operation, were as good as they could be. It turned out later that her predictions of future costs were accurate too. But I found I couldn’t stand up to Eric, at least not in public and not with his calculator. In addition to his intense inquisition of me, I remember being quite open about the uncertainties. I said things like, ‘I don’t exactly know how much this is going to cost, but let’s put an outer envelope on it and see how it goes.’ Truthful it was; but it wasn’t the official way to play the game. Bob put it down to my lack of experience with this kind of review, which was much more like the way they did things in the United States. You really needed to have all the facts and costs at your fingertips, or at least look as if you did. Up till then I’d happily operated with the usual British approach to funding, which Bob characterizes as: ‘If you think I’m a top-notch person and this is a high-class project then you should fund it. As far as the details go, you just have to trust me.’ Bob knew what was needed, but he wasn’t allowed to speak, and must have despaired as he listened to me. But in the event Eric’s interventions and my unfocused responses probably had less influence on the outcome than I imagined.

Later the same day Mike Dexter came to see me and said that the committee was supportive in principle of our doing a third of

the genome, although they were not going to recommend that we be funded to the tune of the full amount that we had asked for. By now the two Sanger Centre backers, the Wellcome Trust and the MRC, were even more unequal partners than they were at the beginning. Our understanding all along was that they would share the costs of the program equally, and the MRC had certainly been participating in the discussions of our future strategy as an equal partner, up to the day of the meeting. What happened was that the Wellcome Trust gave us the money to produce fully finished sequence of one-sixth of the genome by 2002—a half share, as they had promised, about £60 million in total. The MRC, on the other hand, gave us just £10 million over five years, some of which was to finish the worm.

With hindsight, I suppose it was unrealistic to expect that the MRC would come in with equal funds. Almost all of its budget was spoken for, in long-term funding to research units and programs, before it could even think about giving out money for new projects. Diana Dunstan, director of research management at the MRC, confirms this view.

And getting a substantial budget increase from the government was not easy. In fact, following my December 1994 meeting with Dai Rees, the MRC had remarkably quickly won additional funds from the government’s Office of Science and Technology, specifically to support ‘the U.K. arm of the global initiative to generate a sequence map of the human genome in five years.’ It was ironic that the U.K.

government became the first body to put money on the Megalomaniac Genome Project in its original form, although the sum was far short of what was needed. The grant amounted to £2 million per year for five years: in other words, the MRC put nothing more from its own resources into large-scale human sequencing at the Sanger Centre. In addition to the existing worm grant and the new government money (the first two years’ installments for the worm project, all the rest for human sequencing), we had a couple of other MRC grants to David Bentley for human genetics projects, but that was it. The MRC had started the whole field of genomics in the U.K. and had seen the worm sequencing through, but from this point on it became very much a minor player in the Sanger Centre and before long it ceased to be represented on the board of directors of our management company, Genome Research Limited.

Dai Rees, who was then secretary of the MRC, puts the collapse of the partnership down to the inevitable constraints under which a government agency operates.

The Trust did not feel it could make up the shortfall and fund the full third of the genome on its own. Today Mike Dexter, now the director of the Wellcome Trust, supports that decision.

This, of course, was my reason for being vague at the time. The big square buttons and the NIH-style review were meaningless—or, at best, they were a sort of theoretical examination. Nobody could know what the picture would be like three years later, though in fact the guesses that we put forward that day were about right. The initial annual spend would not be all that high because it would be a matter of mapping and starting to ramp up the sequencing, not of jumping straight to full production. By the time of full production in 1998 there might well be better and cheaper technology; but instead of coming to it cold as actually happened, we would be poised and efficient. The best way would be to proceed optimistically and openly, justifying each year as we went along. As it turned out this is what happened in practice, but at the time I failed to put the point across.

Michael Morgan, who unlike Mike Dexter was a Wellcome Trust insider at the time of the review, remembers it differently.

At the time this was the biggest grant they had ever awarded, and more than any sequencing center in the world had been promised. In the absence of any other initiative it sounded very good to say that the Sanger Centre would sequence one-sixth of the entire human genome (at a time when human sequencing had still barely begun). And I suppose I should have seen it as a vote of confidence (although if we had got any less than we did, it would have made a nonsense of the Trust investing so much in the Hinxton site). But to me, and to everyone at the Sanger Centre, winning only half of what we had asked for was a serious setback. It was not enough to drive progress as Bob and I had envisaged.

I remembered the advice John Smith, a pioneer of nucleic acid analysis at the LMB, had given to Alan and me one night in the Frank Lee when we told him that we were going to sequence the worm genome by the end of the decade. ‘That’s too long,’ he said. ‘You can’t make a science project last that long.’ The moral is that if you find yourself in possession of technology that allows you to go at a certain speed and you go at less than the maximum, then you’re done for, because somebody else will come and do it ahead of you. I had made a similar point when the MRC asked what would happen if we got only half the money, and again at the review. But it hadn’t worked. Between us in that review room we had missed our chance, and a couple of years later we were to face the consequences.

Meanwhile in the United States Bob Waterston was having no more luck. It seemed that the national funding bodies, the MRC and the NIH, were of one mind in erring on the side of caution. Bob’s bid for funding to sequence a further third of the human genome was given a doubtful review. The committee had no quarrel with the quality of what Bob was doing; but it argued that it would be a mistake to

spend a lot of money immediately on the existing approach when in three years’ time there might be a better way of doing it. Bob gave a spirited reply, saying that it would be disastrous to delay for three years, but to no avail. He was awarded a quarter of what he had requested in the first year, with modest increases to follow.

Bob shrugged off his disappointment, believing that his detractors were simply postponing the inevitable, as he wrote to me in December.

As in the case of the Sanger Centre, on the face of it he had little to complain about. When Francis Collins announced funding for a series of pilot projects to ‘explore the feasibility of large-scale sequencing of human DNA’ in April 1996, Bob received a third of the $20 million or so awarded for the first year. Five other labs were funded, including Eric Lander’s and Maynard Olson’s. The pilot projects were charged with producing a modest 3 percent of finished human sequence in the first two years; they were to be judged not so much on output as on what they could achieve in terms of accuracy, speed and cost-effectiveness. After a review of these projects in 1998, the funding agency would then decide on a strategy for the final phase.

It was all very admirable in its intention that the ultimate goal should be a complete sequence finished to the highest standards. But it was deeply frustrating to me that the biological community should

have to wait until 2005 for this prize. It wasn’t that I didn’t think the quality of the final product mattered. But I did believe, very strongly, that making a less perfect product available sooner would be better for biology. People could get on with finding genes and understanding how they worked as soon as they had a draft, as we knew from our experience with unfinished worm sequence. And I still think Bob and I were right. There was no miracle new technology. Instead, the tried and tested gel-based approach simply became more and more efficient, costs falling from £0.50 ($0.75) per base to below £0.10 ($0.15) per base in five years, while output from the main centers increased twentyfold. In the light of what happened afterwards—the launch of a private-sector competitor for the Human Genome Project and the subsequent adoption of a ‘working draft’ strategy by the HGP (see chapter 5)—I have no doubt at all that the funding bodies made a big mistake in not supporting us at the time.

Francis Collins thinks the cautious approach was justified, on the grounds that we needed ‘to be sure we knew how to do it at high quality before trying something intermediate.’

As Science reported at the time, the priority of the National Center for Human Genome Research seemed to be to ‘hedge its bets and circulate its limited funds to a variety of labs.’ Some outcomes were positive:

Eric Lander succeeded in getting his highly automated shotgun production lines running, and this later paid off for him in generating draft sequence quickly. But time was lost in getting on with the job: the public side was seen to be slow, and this subsequently made for a weak PR position. ‘It would have been much better to have accepted John and Bob’s idea at the time,’ says Jim Watson now.

Even though we were not able to expand our sequencing effort as fast as we wanted, human sequencing really began to take off in 1995–6. As Bob and I had been hatching our megalomaniac plans, we looked at what was happening around the world and recognized that it was pretty chaotic. People were duplicating what others were doing; everybody was piling into regions of the genome reputed to have genes for cancer or other major diseases; there was no consensus on whether or how the sequence data should be made available to the community. If your goal is a complete genome sequence you can either do it yourself and ignore the rest, which means you have to go faster than everyone else, and you have to succeed; or you have to get everybody together. To a certain extent Bob favored the first option; he really thought we could split the whole thing half and half and do it all. But I’ve always felt that if you can share you should do so—that way you don’t have silly competitions.

I also felt that there was a need to strike a blow against the gold-rush mentality that was developing towards the genome. There was a sense, largely but not exclusively fostered by the new breed of genome-based private companies, that everyone was in a race to stake claims as fast as they could and reap huge profits from their discoveries. A case in point was the public-versus-private drama that was beginning to unfold at about the same time over the breast cancer genes, a drama in which the Sanger Centre was playing a supporting role.

Mike Stratton, who now runs the Cancer Genome group at the

Sanger Centre, was then leading a team at the Institute of Cancer Research in Sutton, Surrey, dedicated to finding genes that place women at high risk of developing breast cancer. One gene, BRCA1, had been located by Mary-Claire King in the United States. But it could not account for all families with clearly inherited susceptibility to the disease, so Mike and his colleagues set out to find another. In the summer of 1994 they located the gene, which became known as BRCA2, on chromosome 13. Up to this point Mike had been collaborating with Mark Skolnick at the University of Utah, who was cross-checking the massive genealogical database compiled by the Mormons against cancer registries in order to identify families that could help in tracking cancer genes. Skolnick had set up a private company called Myriad Genetics (see p. 110) specifically to go looking for cancer genes and to market genetic tests for any that they found. Not long before his own group located BRCA2, Mike went out to talk to Skolnick to find out exactly what Myriad’s role would be if the collaboration did indeed find and clone the BRCA2 gene. The answer was that Myriad would patent it and own exclusive rights to exploit it both for diagnosis and therapy. Mike was very much opposed to this idea.

Mike and his colleagues were now racing the Utah lab to find the gene itself and clone it. At that point the Sanger Centre became involved, when they came to ask David Bentley if he would make a clone map covering the approximately 1 megabase region where they knew the gene lay. He was happy to do that. But, having got the

clones, David realized that the BRCA2 region would make a nice pilot project for the human sequencing program at the Sanger Centre and Washington University. He asked Mike what he thought. The sequence would be very valuable; but at the same time, the Sanger/Washington University data release policy meant that as soon as it was complete it would be publicly released, and that might help Mike’s competitors. David was very clear that we would not sequence the region if the Institute of Cancer Research team were unhappy about it. ‘We discussed it,’ says Mike ‘and decided that as we were in this game to get the gene found, it would be ridiculous for us to even consider not allowing David to go ahead.’

David was soon able to give Mike a date, November 23, 1995, when the sequence would be placed in the public databases. Shortly before that, the ICR team found a mutation from one of their breast cancer families that looked as though it might very well sit in the BRCA2 gene. Within two weeks of the sequence being available, they not only confirmed this mutation but found five more. There was now no doubt: they had found the gene. Mike moved fast to publish the group’s discovery in Nature, while keeping it secret even from his collaborators until the last possible minute. But despite his efforts, enough information about the discovery reached Skolnick to enable him to locate the gene himself and bang in a patent application—the day before the ICR paper came out in Nature.

Despite his reservations about patenting, Mike had realized that in such a competitive area, the only way to protect his team’s discovery from commercial exploitation by others was to patent it himself. The Institute took out one patent on the first mutation as soon as it was discovered, and another later covering more mutations. Meanwhile Myriad’s patent applications claimed rights to the whole gene. The Utah scientists had been the first to clone BRCA1, on which they also own patents. They set up a commercial diagnostics center in Utah and, once the patents were granted,

threatened legal challenges to any lab elsewhere in the United States that was using either gene to carry out breast cancer screening. All such screens henceforth had to be done at their own center, at a cost of around $2,500 per patient. Other labs could apply for licenses to carry out simpler tests to look for single mutations, but again had to pay a fee of a couple of hundred dollars per test. One of these tests is for a mutation in BRCA2 found by Mike’s team, which is particularly common in the Ashkenazi Jewish population (those who originated from Central and Eastern Europe). ‘The Ashkenazi BRCA2 mutation was in our original paper,’ says Mike, ‘so Myriad is claiming a fee from all women who undergo tests in the United States for a mutation that was discovered by us.’ As an Ashkenazi Jew himself, Mike found this particularly hard to take.

In Mike’s view it is ‘unfair and unethical’ that Myriad should have complete control of screening for breast cancer susceptibility, especially when (as Myriad acknowledged in its 1996 paper) their findings on BRCA2 depended on the earlier work of Mike’s team. Only the Institute of Cancer Research patents stand in their way, and in fact Myriad has been less successful in Europe than it was in the United States at challenging the use of the gene. But having accepted massive financial investments, the company now has no alternative but to market its goods as aggressively as possible. And as a body largely funded by coins dropped in the tins rattled by Cancer Research Campaign volunteers, the ICR cannot justify spending the huge sums on lawyers that would be needed to fight Myriad through the courts.

I see this as a cautionary tale. In my opinion commercial imperatives have overridden clinical and ethical imperatives, just as Mike feared. By claiming rights to diagnostic tests for the two BRCA genes and charging for them, Myriad adds to the total bill for health care. In the United States the tab gets picked up by the patient, through higher insurance premiums; in the U.K. it’s the taxpayer who foots the bill through increased costs in the National Health

Service. But much worse than this is the impact on science and future treatments. Once scientists really understand how the mutations in BRCA1 and 2 unleash the growth of tumors, they might be able to devise new therapies. But because of its patents, only Myriad has the right to market such therapies. Others with relevant expertise will have less incentive to join the effort; even if they make new discoveries there will be a lot of work for lawyers involved in sorting out cross-licensing agreements. The total amount of brainpower focused on the problem will be less as a result. It seems to me that companies such as Myriad are going for short-term profits at the expense of long-term benefits to human health—the promised benefits that are the ultimate justification for the whole genome sequencing enterprise.

Although in 1995–6 the full consequences of Myriad’s aggressive approach had yet to emerge, it was pretty obvious where a focus on commercial profit would lead. Bob Waterston and I felt there was a need to get some kind of commitment from the international sequencing community that genomic information would be made publicly available and not parcelled out among the claims of the profit-mongers, with individual deals between companies and researchers. Michael Morgan, who in 1994 had been deeply involved in discussions in HUGO and elsewhere about the problem of open access to ESTs, was confident that the Wellcome Trust could play a role in bringing this about. He, Bob and I came up with the idea that what was needed was an international meeting to hammer out a strategy for deciding who would do what, and how the data would be managed. Michael got Bridget’s agreement that the Trust should sponsor the meeting. In discussion with the other main funding bodies, the National Institutes of Health and the United States Department of Energy, he settled on Bermuda as a suitably neutral location (British territory but within easy reach of the United States), and the date was fixed for the end of February 1996. We began to draw up a list of those who should be invited—anyone who looked

as though they were serious about genomic sequencing (and had funding) as opposed to those who were simply sequencing genes. Once they heard what was going on, others started banging on the door to be allowed in. The list included representatives from the Department of Energy sequencing labs, Mark Adams and Craig Venter from TIGR, Eric Lander from the Whitehead Institute, many others from United States labs, sequencers from the U.K., France, Germany, Italy and Japan, as well as representatives of the databases and the funding bodies, and individuals who had influenced genome policy.

The meeting, held in the Hamilton Princess Hotel under grey skies in the off season, turned out to be extremely constructive. It was the first opportunity people had had to compare notes in a comprehensive way. I think it did an awful lot of good; simply talking about the amount of sequencing that Bob’s lab in St. Louis and the Sanger Centre were doing had quite a salutary effect. Those who were sequencing a few thousand bases could no longer claim to be major players, and anyone pursuing a less effective strategy was immediately exposed: the meeting had a major impact in consolidating strategies and methods. The smaller groups realized that they had either to get grants and reorganize themselves to do a significant chunk, or to accept that they were going to be gene sequencers now, and not part of the big thing. The most important thing about that first Bermuda meeting was its political aspect, sorting out who was doing what. It was there that we first began to work out what I called the ‘etiquette of sharing.’ We had to work together, because nobody at that stage could do the whole thing by themselves. Everyone arrived with claims on pieces of paper announcing their intention to sequence a particular region, and during the course of the meeting any competing claims were sorted out. One of the outcomes of the conference was that all these claims were recorded on a website called the Human Sequencing and Mapping Index, which was initially managed by Susan Wallace, administrator of the

United States office of HUGO. It was a logical extension of the role HUGO had established for itself of keeping track of who was doing what in the genome.

At the end of the meeting we had a session to discuss the question of data release, and this is the issue with which the Bermuda meeting is most associated in people’s minds. Bob and I had always released the worm data promptly, and we continued to do the same with human sequence. But we were aware that not everyone did. There was no mechanism at the time for putting unfinished data in the public databases; they were at the time for finished sequence only. But, just as we had with the worm, we made all of our unfinished human sequence data available electronically from our own sites, so that anyone could download the information and do what they liked with it. We just asked that they recognize that it was unfinished, and acknowledge the source of the data in any publications. Everyone at the Sanger Centre, and anyone who collaborated with us, had to accept that we would do this. It was something we were doing all the time, often against opposition from those who wanted to find useful (and maybe profitable) genes before anyone else did.

I felt strongly that the principle of free data release had to be accepted, or nobody would trust anyone else. Bob and I were running the session in Bermuda, and I found myself standing there in front of a horseshoe of chairs, making my pitch. I thought it pretty unlikely that everyone would agree; several of those present, who included Craig Venter of TIGR, already had links to commercial organizations and might oppose the idea of giving everything away for nothing. But as I stood there, scribbling away on the white board, rubbing words out and rewriting, we hashed out a statement. The Wellcome Trust still has a photo of that handwritten statement, with its three bullet points. It reads:

• Automate release of sequence assemblies > 1kb (preferably daily)

• Immediate submission of finished annotated sequence

• Aim to have all sequence freely available and in the public domain for both research and development, in order to maximize its benefit to society

While Bob and I were working with our scientific colleagues, Michael Morgan was doing the same with the representatives of the funding agencies.

What I had written on the board, with minor modifications, became known as the Bermuda Principles, which have served as a point of reference for publicly funded large-scale sequencing ever since. I was amazed that in the end everyone put their hands up to this; I had no idea that it was going to go so far. Some of the delegates agreed with the principles but had difficulties with their practical implementation because their national governments were strongly opposed to the requirement for rapid release, which implied that the sequence could not be patented. ‘A lot of the smaller countries did not trust the United States,’ says Michael. ‘They thought they were just pretending to be committed to free release while patenting on the side.’ But there could be no exceptions, otherwise the whole thing would break down. Open access and early release mean that anyone in the worldwide biological community can use those data and turn them into biological understanding and ultimately into new inventions that can be patented. But the sequence itself in its raw form when publicly released becomes unpatentable. And in Bermuda, for the first time, we won the acceptance of most (though not all) of the genome sequencing community that this was a desirable state of affairs. It

boded well that so many people had come to share the same vision of the genome sequence as ‘the heritage of humanity’, the phrase adopted in the first article of the Universal Declaration on Human Rights and the Human Genome at the General Conference of UNESCO in 1997.

The Bermuda conferences, or to give them their proper name the International Strategy Meetings, became annual events. Although after 1998 they were no longer held in Bermuda, they continued to have the important dual purpose of regulating who claimed what as their sequencing territory on the genome, and maintaining the data release policy. They also began to set standards for the quality of the data that were released.

By the middle of 1996 the scene was really set for a concerted attack on the human genome. There was an international consortium, more or less in agreement and with serious levels of funding (although not as much as Bob and I would have liked). At the Sanger Centre we had teams set up, and a plan for how we would progressively increase our output of human sequence. In July we moved into our new building, gleaming and futuristic (at least compared with the old brick buildings we’d been in so far), set among trees and next to an artificial lake. Although the increased space was welcome, we immediately encountered a problem that I characterized as adiabatic expansion. What happens to a compressed gas when you give it more space? It gets cold. In the old buildings we had all been more or less on top of each other, with many of the sequencing machines in a central area nicknamed the ‘goldfish bowl’ surrounded by offices off a raised gallery. In the new building we all lost touch with each other as we disappeared into labs and offices separated by miles of corridors. I’m very dependent on informality, and was rather unhappy that I would not be bumping into people all the time any more as I had in our previous cramped quarters. We had to meet more regularly, to make sure we kept in touch. But I

was confident that everyone knew what he or she was doing.

Later the same year the Sanger Centre shared in the celebration of another major landmark: the completion of the yeast genome, which was published in October. An international (and largely European) consortium in which Bart Barrell’s group at the Sanger Centre and Bob’s colleagues at Washington University had played a major role, the yeast sequencing project was the first to decipher the genome of an organism more complex than a bacterium. Apart from providing an estimate of the number of genes needed to run a cell with a nucleus, similar to those of higher animals (an article in Science was headed ‘Life with 6000 genes’) and opening the way to understanding many of the functions of such cells, the yeast sequence was a huge asset to the worm sequencing project. We could now assemble the sequence of our YAC clones efficiently because we could eliminate the sequences that belonged to the yeast. With the yeast genome all in the databases, it was now just a matter of cross-checking electronically. Nothing, I thought, could now stand in the way of getting the worm genome finished.