2 Unsung Contributors to the Human Genome Project

In response to ongoing debates within the biological research community in the late 1980s about a proposal to map and sequence the human genome, the Board on Basic Biology of the Commission on Life Sciences of the National Research Council (NRC) appointed a special committee “to examine the desirability and feasibility of [such a project] and to suggest options for implementing the project if it were deemed feasible” (NRC 1988). In 1988, the committee issued a report that began “A special effort . . . should be organized and funded specifically” to acquire “a map, a sequence, and an increased understanding of the human genome . . .” (NRC 1988).

The report’s Executive Summary began by noting that “Humans have long been intrigued by the forces that shape them and other organisms”; that a century of biological studies had led to the conclusion that, in the humans as well as all other animals and plants, inherited traits are controlled by genes; that genes are composed of DNA; and that “as more of our genes are mapped and their DNA sequenced, we will have an increasingly useful resource [emphasis added]—an essential data base that will facilitate research in biochemistry, physiology, and medicine.” To that end, the committee concluded that a “special effort” should begin immediately to construct detailed genetic and physical maps of the human genome and the genomes of important model organisms, and to mount a “focused effort” to develop the “required advanced DNA technologies” to analyze genomic DNA. The committee recommended that this be supported with additional funds, “not diverted from the federal research budget for biomedical sciences,” that would rapidly scale up to $200M per year. Significantly, the report noted that this “human genome project should differ from present ongoing research inasmuch as the component subprojects should” have quantitative goals and progress be assessed quantitatively, and “should have the potential to improve by 5- to 10-fold increments the scale or efficiency of mapping, sequencing, analyzing, or interpreting the biological significance of the information in the human genome.” It was predicted that the special effort could be carried out “in the next two decades.”

While there was a wide-ranging set of responses to the NRC’s recommendations, from vehement disagreement to enthusiastic support, the National Institutes of Health (NIH) and the Department of Energy (DOE) in the United States, and agencies in several other nations, accepted the recommendations and the Human Genome Project (HGP) officially began on October 1, 1990.1 Less than thirteen years later, the International Human Genome Consortium announced (International Human Genome Sequencing Consortium 2004) that a highly accurate, highly continuous, “essentially complete” human genome sequence had been finished. In the course of the last decade of the twentieth century and the first three years of the twenty-first century, the public human genome effort generated:

• a sequence of “2.85 billion nucleotides interrupted by only 341 gaps [covering] ca. 99% of the euchromatic genome [at an accuracy of] ca. 1 error in 100,000 bases (99.99%).”
• The total cost of this sequence was less than $0.09 per finished base, a reduction in cost of at least five orders of magnitude.
• Other accomplishments included a 1 centimorgan human genetic map (Gyapay et al. 1994), and maps of 3.7 million human single-nucleotide polymorphisms (International HapMap Consortium 2005) and 15,000 full-length human cDNAs (Mammalian Gene Collection [MGC] Program Team) (Mammalian Gene Collection 2002), plus the complete DNA sequences of E. coli (Blattner 1997), S. cerevisiae (baker’s yeast) (Mewes et al. 1997), C. elegans (a roundworm) (C. elegans Sequencing Consortium 1998), D. melanogaster (the fruit fly) (Adams et al. 2000), and a draft sequence of M. musculus (the laboratory mouse) (Mouse Genome Sequencing Consortium 2002).
• The project also developed many disruptive technologies, in addition to DNA sequencing methodology.
• The project’s embedded program in the Ethical, Legal, and Social Implications (ELSI) of genomic research became a major funder of bioethical studies in the United States.

The fact that the HGP was completed at all, not to mention completed at a cost less than originally estimated and in a time frame faster than originally predicted, was a stunning achievement, even to those of us who had been deeply involved from the beginning. It was an event of enormous satisfaction and tremendous pride to all the participants (as well as of great exhaustion to many), and one that was universally recognized as being a true game changer for all aspects of biological and biomedical research. So, how did the participants manage to pull it off?

The data and analyses comprising this amazing historical achievement were, of course, generated through a massive, sustained effort of scores of laboratories and hundreds, if not thousands, of individual researchers, whose approaches, failures, and successes have been well documented (Davies 2002). Less well documented are the contributions of the many scientist-administrators who guided the participating funding agencies’ efforts and, very importantly, coordinated each agency’s activities with those of the other funders. Another critical set of contributions came from the above-the-call-of-duty participation of the scientific advisors upon whose wisdom and judgment the active participants relied; this critical outside input started with the NRC report and continued throughout the entire thirteen years of the project. The purpose of this paper is to describe the activities of those participants and some of the lessons learned in the course of bringing the HGP to its successful completion.

From its outset as Elke Jordan vividly describes in her contribution to the volume, the HGP posed many challenges, the primary one being that it was an audacious effort, of a scale that had never previously been attempted in biological research. And it had to be started in the face of a significant amount of controversy in the scientific community (Weinberg 1988; Angier 1990) Some scientists, while recognizing the value of the planned resource, did not think it could be done in the planned time frame or for the proposed cost. Others thought that, even if it could be completed, the resource that would be generated would be of little scientific value and would not add significantly to the efforts of their own research activities. The critics saw the proposed HGP as a serious misdirection of the government’s funds for support of scientific research (Angier 1990).

While acutely aware of those objections and in some cases taking them into account, the participating scientists and involved NIH staff committed to achieving the long-term goals and pressed on into the challenging unknown, betting the future of their careers on the value of what they were trying to achieve. Implementation was conceived in both the short (first five years) and long (fifteen-year) terms, and was implemented according to plans that were, throughout the project, developed in series of discussions and meetings, which very consciously included skeptics among the attendees (see for example Collins et al. 1998)

Even in the face of warnings against participating in the HGP at such a vulnerable point in their careers, given the high-risk nature of the project and unsure future funding, a disciplinarily diverse set of young scientists, who had not previously worked together, was recruited to begin the work. The required scale, novelty, and rapid pace of the project presented many challenges, both conceptual and technological, to them as well as to the NIH2 staff. For the latter, one of the challenges was to develop ways to support high-risk/high-payoff projects, which meant that failures of a large number of the individual efforts needed to be anticipated.

Staff had to develop mechanisms (discussed below) that would allow rapid response both to such failures and, probably more importantly, to significant successes or changes in technologies. A particular challenge was disruptive developments, innovations that fundamentally changed the ways in which the project was being carried out. Both disruptive successes and significant failures required individuals and the overall project to redirect efforts and/or funding more rapidly than the NIH grant system usually allowed. Dealing with such issues regularly could have, over time, discouraged the scientists and staff working on the project. However, they continued to be motivated by their continuing belief in the ultimate value of the project and their conviction that it was worth trying to push to do it in the time frame and with the budget proposed. They were confident that once data production began and the data became widely available, the criticisms would disappear as it became clear to the scientific community that the HGP was a value-generating activity, and that the resource generated would transform biology.

The HGP can be thought of as (1) a large data production effort consisting of two phases, mapping (both genetic and physical) in the first and sequencing in the second and (2) associated efforts in technology development, bioinformatics, pilot sequencing projects to inform the strategies and methods for the human sequencing effort, and a program to support research about the ethical, legal, and social issues raised genomics. To deal with the complexity attendant in initiating and maintaining these various components, and providing the necessary degree of integration, the NIH relied heavily on input from outside experts. For example, in 1990, the official start of the HGP, the NIH and DOE presented their joint plan for the first five years, “Understanding Our Genetic Inheritance. The U.S. Human Genome Project: The First Five Years, FY 1991–1995” (National Center for Human Genome Research 1990). This was written by the agencies’ staff, with input from three outside committees—a Joint Mapping Working Group (six members); a Joint Informatics Task Force (fourteen members plus ten additional contributors); a Joint Working Group on Ethical, Legal, and Social Issues (six members); plus six other senior scientists. The involvement of these forty-plus outside scientists set a precedent for the three subsequent five-year plans 1993, 1998, and 2003 (see Collins et al. 2003), each of which had input from a couple of hundred workshop and meeting attendees.

The NCHGR, and later the NHGRI, also got required input from the National Advisory Council for Human Genome Research (NACHGR), whose formal approval was required by law before any grant solicitation could be issued or any funding decision could be made. The Council, which met regularly three times per year, also provided advice about other agency activities, and the overall progress of the NIH’s contribution to the Human Genome Project. The DOE got similar input from its Health and Environmental Advisory Committee (HERAC). The two agencies’ interactions were overseen by a Joint DOE-NIH Subcommittee on the Human Genome. Finally, as the program at NIH progressed, a number of individual committees of external scientists were convened to provide more specific, detailed guidance for many of the individual programs, including sequencing.

A more specific example of how the NCHGR/NHGRI used outside advice in 1998 concerns events in 1998 when an updated version of the NIH/DOE strategic plan was needed. A meeting of about 200 participants was held at Airlie, a conference center in Warrenton, Virginia, to discuss the new plan. NIH staff developed a draft based on input obtained from several individual workshops on specific issues, as the basis for the meeting’s discussions. Upon its presentation in the opening session, however, many of the attendees found serious fault with the draft, and the agenda of the meeting was revised on the fly to allow consideration of their concerns. By the end of the three-day meeting, a restructured, much-improved draft was produced, which went on to be approved by both the NACHGR and the HERAC. Over the thirteen-year lifetime of the HGP at NIH, the advice that was solicited and obtained from several hundred outside experts was critical to the advancement and ultimate success of the program.

At the NIH, the unique activities (and controversial nature) of the human genome effort demanded some innovative administrative measures. A new funding agency, the National Center for Human Genome Research (NCHGR) (which later became the National Human Genome Institute [NHGRI]),3 was created to ensure that the Congress was able to appropriate funds that were (1) specifically used for genomic research and (2) clearly not taken from other research activities at the NIH. Because of the nature of the research that would be required to carry out the HGP (e.g., research designed to meet specific production goals, high-risk/high-payoff research, hypothesis-generating research as opposed to hypothesis-testing research), new review committees had to be established, and new review criteria had to be developed. Program staff found that they needed to use existing funding approaches in innovative, non-standard ways, which often required obtaining permission from the appropriate NIH offices (see later discussion of cooperative agreements).

An early example of the way in which new thinking had to be applied to implement the HGP at NIH came with the strategic decision to support large-scale genetic and physical mapping on the basis of individual human chromosomes. Staff first proposed to use a Centers mechanism (P30 and P50 in technical terms), but several Council members did not initially sanction that approach. They were concerned about committing large amounts of funding in a small number of places. Those concerns had to do with the potential difficulty of terminating such large grants or otherwise holding investigators accountable if they failed to deliver, and were based on the perception of some of the Council members that the funding for some NIH Centers in other fields had continued long after their useful lifetime. After debate about other funding mechanisms, including contracts, the initial mapping efforts were funded as program projects (P01).

But early experience revealed problems with using the P01 mechanism for these purposes. Program projects consist of related, but quasi-independent, research projects led by different investigators, each of whom has the inherent autonomy to pursue their projects in different directions. The HGP mapping efforts, however, needed much more integration and singleness of purpose. Accordingly, the Council subsequently approved the use of P30 and P50 grants to fund Genome Science and Technology Centers (GESTECs), but only after obtaining assurance from staff that these awards would be very closely monitored, and renewals would only be funded with Council approval after a clear and convincing demonstration of a continuing need. Staff achieved the close monitoring expected by exercising greater than normal interactions with the investigators, expecting more frequent reporting, having regular telephone calls to discuss progress and conducting yearly site visits. Over the course of the first few years of the HGP at the NIH, eighteen GESTECs were supported for a wide range of genomic activities, including integrated genetic and physical maps of human chromosomes 3, 5, 6, 7, 11, 12, 13, 15, 17, 22, and X; development of three different sequencing technologies; sequencing of the nematode and fruit fly genomes, as well as S. cerevisiae chromosomes IV (half) and V; genome-wide mapping of the human and mouse genomes; and new technology for identifying disease genes.

The rapid development of human genetic maps is a good illustration of the rapidity with which progress was made in the HGP and the flexibility required on the part of NIH staff to respond to that progress. The first complete human linkage map was published in 1987 (Donis-Keller et al. 1987) before the HGP began, and comprised 403 restriction fragment length polymorphism (RFLP) markers. Initial NIH genome funding included support for efforts to improve the maps of individual chromosomes. However, RFLP mapping was relatively slow and laborious, and by 1991, questions were already being raised about whether the HGP human genetic mapping goal of a 1 centimorgan (cM) map would be achieved. In response, NCHGR created an “index mapping project” to rapidly construct relatively low-resolution (10–15 cM) maps composed of very highly polymorphic markers to provide an interim resource for the human genetics community, and to create scaffolds for building the higher resolution maps called for by the HGP. Shortly thereafter, however, the index mapping effort was overtaken by a whole-genome mapping effort made possible by the development of a new type of genetic marker, the microsatellite repeat (or short tandem repeat polymorphism, or simple sequence length polymorphism). The advantages of microsatellites over RFLPs were quickly recognized and the genetic mapping field, including NHGRI-supported investigators, largely abandoned RFLP mapping and switched to microsatellite-based mapping. In 1991, NHGRI distributed a list of 283 highly informative human genetic markers, of which 163 were microsatellites. In 1994, the French research group at Genethon published a linkage map of 2,066 microsatellites, which had been constructed on a genome-wide, rather than individual chromosome, basis (Gyapay et al. 1994). Finally, a 0.7-cM human linkage map, exceeding the HGP goal, was developed by the Genethon group in collaboration with three NHGRI-supported laboratories and one other French group. This map comprised 5,826 loci (3,617 microsatellites, 427 genes, and 1,788 other polymorphisms.) The NHGRI then pulled back funding for genetic mapping and allowed (and encouraged) its grantees to turn their attention to physical mapping (Murray et al. 1994).

The human genetic mapping effort laid the groundwork for future genome activities in terms of being collaboratively generated, international in scope, based on a new technology that supplanted existing technology, and integrating all available data into a new, useful resource available to the entire scientific community. The new genetic map effectively replaced the gene-by-gene or trait-by-trait approach that had previously been the norm in human genetics. Investigators now had a better than 99 percent chance of localizing a gene of interest near a mapped marker, within a region of interest less than 0.2 percent of the human genome. This did mean that some investigators had to abandon the methods they had used for some time and learn to take advantage of the new resource, but it clearly demonstrated the overall value of the first product of the HGP to the larger research community, and exemplified the promise of the other resources, which the HGP intended to generate.

There were important lessons for both investigators and the NIH staff in the success of the HGP’s linkage mapping phase. In place of the intense competition that had characterized the field of human genetics up to that time, the cooperation that was encouraged and that developed among the HGP’s participating scientists was productive as well as rewarding. Staff, too, had to make changes. Having come from an administrative background that emphasized the success of individual grants and grantees, we had to learn new ways of encouraging and managing what came, in time, to be recognized as “team science.” From the beginning of the HGP, the investigators and administrators had to learn on the fly to develop powerful new approaches to achieve goals that had seemed to many to be futuristic, even impossible. One of us likened the atmosphere to that of a start-up biotech company in that the effort was very much in the public eye, that every decision seemed to be absolutely critical for the future of the enterprise, and that it was in the interest of each participant for the others to be successful. And for both the investigators and the administrators, it was critical that they develop new modes of working with others to further the overall project, but to do so in a way that would still allow the investigators to be competitive at the next grant renewal. The line between cooperation and competition was a fine one that required constant attention throughout the HGP.

The remaking of the field of human linkage mapping by the invention of a type of genetic marker that could be assayed by a common method, polymerase chain reaction (PCR), and was easily transferable from the lab that developed it to all other labs was not lost on those engaged in the physical mapping efforts of the HGP. Prior to starting the HGP, physical mapping was a broad term encompassing different types of methods for ordering pieces of DNA, such as in situ hybridization mapping, deletion mapping, long-range restriction mapping, and clone overlapping. In 1989, Olson et al. (1989) introduced the Sequence-Tagged Site (STS) as a “common language” for physical mapping. Since this element was defined only by its unique DNA sequence and could be assayed by PCR, STSs could be ordered in any of several ways, and individual STSs provided a connection between physical maps constructed by different techniques (polymorphic STSs also allowed connections to be made between the physical and genetic maps of a given region). Although, once again, there was considerable skepticism about its value when the STS concept was first described, it was quickly adopted by the HGP, which then defined its physical mapping goal as STSs spaced at intervals of approximately 100 kilobases (100,000 bases) and overlapping clone sets that represented physically contiguous regions (contigs) of the genome.

The physical mapping component of the HGP was, in some ways, the most difficult and most problematic. While clone/STS maps of some human chromosomes were published in the early 1990s (see for example Green and Green 1991; Collins et al. 1992), most if not all of the physical mapping strategies had some significant problems that had to be dealt with. Overall progress in physical mapping was more difficult to track than for either genetic mapping or sequencing. The capacity of some cloning vectors, such as cosmids, was too small to be practical to use in the construction of a clone-based map of the entire human genome. Other vectors, such as yeast artificial chromosomes, that could incorporate very large pieces of chromosomal DNA (up to a few megabases), yielded libraries that were frequently unstable (the inserted DNA suffered deletions as the clones were grown) or chimeric (DNA regions from different parts of the human genome were artefactually joined). Some regions of the human genome (e.g., highly repeated regions, centromeres, telomeres) did not seem possible to clone, at least in the available vectors. Because of these and other problems, quality control became a significant issue in physical mapping. In 1994, an international group of scientists, including several NHGRI grantees, proposed a set of standards for reporting physical maps (National Center for Human Genome Research 1994). With the advent of sequencing, physical mapping became integrated into the sequencing pipeline, with the sequencers collaborating with mappers to obtain mapped clones or building the physical maps as they went along. However, whatever the deficiencies were in the effort to provide quality control for physical mapping, the need for quality control was established and then became a critical feature of the sequencing phase of the HGP.

Genomic DNA sequencing began at the NIH, as recommended in the 1988 National Academy of Sciences report, with efforts to sequence smaller genomes. The international public effort focused on the sequencing of the genomes of a small number of important model organisms. Each of these involved a different organizational and technical strategy. The E. coli sequencing effort (Blattner et al. 1997) was pursued by a single laboratory using rather standard, gel-based sequencing methods, and was funded entirely by the NHGRI. The S. cerevisiae genome was sequenced in a distributed, chromosome-by-chromosome basis by a large number of investigators separately funded primarily by European sources, with some (<15 percent) participation by the NHGRI (Engel et al. 2014). The C. elegans genome (C. elegans Sequence Consortium 1998) was sequenced by a two-laboratory collaboration that used and, in the course of their efforts, dramatically improved capillary-based sequencing; that project was jointly funded by the NHGRI and the UK Biomedical Research Council. This effort was widely supported by the C. elegans research community, which had a tradition of openness and sharing. The sequencing of the D. melanogaster genome (Adams et al. 2000), which was carried out in a single laboratory and supported by the NIH, and the DOE in contrast did not have widespread support in the fruit fly research community until the data were generated and released, when the utility of the sequence data was recognized by most researchers. The sequences of other organisms were determined by others and also made an important contribution to the dawning age of genomic sequencing. A whole genome shotgun sequencing strategy, for example, was used in determining the first sequence of a microbial genome, that of Hemophilus influenzae (Fleischmann et al. 1995)

The effort to sequence the human genome began sooner than almost everyone, especially the NHGRI staff, thought that it could, would, or should. In 1994, John Sulston and Bob Waterston, the principal investigators of the two labs (at the Sanger Center and Washington University, respectively) sequencing the C. elegans genome, approached James Watson (Ferry and Sulston 2010) and then NIH with a proposal to start large-scale sequencing of human DNA. This was at least two or three years before NIH had expected to start funding such sequencing efforts.

The NIH response to the proposal is a good illustration of how the agency staff responded to such disruptive developments. A meeting of DNA sequencing experts was rapidly organized. After intensive discussion, the attendees of that meeting agreed to support the proposal and urged the NIH to start funding human sequencing at a higher funding level. Based on advice from advisory committees, the NIH, however, did not immediately go full tilt into human sequencing and instead pursued a middle ground, probably to the dismay of those investigators who had advocated for a larger scale effort to begin. In 1995, the staff wrote and NHGRI issued a solicitation for proposals for pilot projects to sequence megabase regions of the human genome, as well as to address the difficult issues of obtaining an adequate supply of mapped clones for the sequencing, new technology, sequence quality, sequencing costs, and an ability to scale up sequencing during the three-year period of the grants.4 It had become clear to staff by this time that questions pertaining to all of these issues would need to be answered in order for the staff and advisors to consider scaling up the project in 1999 after the end of the pilot project period.

Several important administrative changes occurred during this period. At the request of staff, quantitative sequencing goals were set by investigators for each year of their awards. NHGRI staff began to actively plan how to move more funds into the large-scale sequencing program by 1999 to enable the program, if successful, to scale up dramatically at that time (described below). It then turned out that the investigators counted and reported completed sequence in many different ways. So, NHGRI had to develop a standardized, quantitative progress report, which was implemented along with a strong urging that investigators use it. A policy for data release was developed and timelines for deposition of data in the public databases were established. Yearly administrative reviews, including site visits by advisors and staff, were conducted. A management plan for the overall sequencing program was developed, and individual management plans, which had to be approved by staff, were requested from each sequencing center. Quality control exercises were started among the centers to assess the accuracy of sequence data and assemblies. And a new, activist sequencing advisory committee was assembled; this committee regularly reviewed progress reports and production figures from, and made regular site visits to, each sequencing center. NHGRI staff took the lead in all of these initiatives, as more explicitly described below, for some.

One interesting, potentially confounding issue that occurred late in the pilot project period was the realization by NHGRI staff that the donors of the DNA being used in the sequencing of cloned libraries had not been properly consented. Staff quickly examined the ethical issues of the problem, were able to get interim approval for a few of the libraries to be used by the sequencers while new, properly consented donors were recruited, and new libraries were made.

Even in this pilot period, the rate of mapping fell behind the rate of sequencing, and clone supply was a continual issue. One consequence of this was there were problems with mappers and sequencers “claiming” genomic territories to work on which others were discouraged from working on. Staff had to help address this by developing a new set of rules about “claiming” territory and ceding previously claimed territory if overall progress was being slowed. During the subsequent production phase, the sequencers changed to a whole genome shotgun sequencing strategy, which did not require mapping. This critical development is discussed extensively in Ferry and Sulston (2010).

Before that change, however, the public sector HGP received a serious challenge. By the end of the second year of the pilot projects, 22.8 MB of finished DNA sequence had been deposited in the GenBank. This was a substantial amount of new data compared to what had been available previously, but still was only about 5 percent of the total human genome sequence. In spring of 1998, representatives of a for-profit biotechnology company met with the sequencing center investigators and NHGRI staff to discuss its plans to sequence the entire human genome with a different approach than the clone-by-clone, chromosome-based, finished sequence approach being pursued by the public sector effort. The company’s plan was to perform random “shotgun” sequencing of the entire genome (all the chromosomes at once) and rely on reassembly algorithms to put the huge jigsaw puzzle of sequences back together and generate the sequence of the entire genome. Further, the company planned to sell the sequence and assembled genome data to the scientific community, bypassing public deposition. The company estimated that it would be able to offer a complete genome to the scientific community by 2001.

This announcement caught the public project off guard and stimulated many discussions and proposals as to how it should react. The first order of business was to assess the quality, contiguity and value to the community of a sequence assembled by the two strategies. A “Summer Sequencing Project” was thus conducted by the sequencing centers, coordinated by the NHGRI staff. It involved each of the sequencing centers sequencing a genomic region of its choice by both strategies and comparing the results. At the end of the project, the participating investigators agreed that a substantial increase in the rate of sequencing the human genome could be accomplished by accelerating raw sequencing capacity, but the product would only be a “high quality” draft version. Biologists who tested the draft-level sequence data found them to be relatively useful for their studies, but not as useful as the sequence that the HGP had promised (see NHGRI Internal History Archive. Scanned Box List 6, File 23 PI Meeting 9-3-98). Almost all participants concluded that having a draft quality sequence sooner would be valuable, as long as the commitment to a fully finished sequence was maintained. Accordingly, the public project published a new five-year plan in October 1998 announcing the sequencing goals for the project were being modified to accomplish a “working draft” of the genome by 2001 and a finished, high-quality genome sequence by 2003, two years earlier than the original goal. There would be no change in the public effort’s data release policy; it would continue to rapidly deposit all its data in the public domain, where it would be “totally and freely accessible.”

This modification of the HGP plan came at the same time as the NHGRI sequencing pilot projects were coming to the end of their grants. The complexities of navigating this period in the HGP can be seen in the difficulties that program staff had in writing the RFA for the full-production phase of the sequencing effort. The initial RFA soliciting applications for scale-up projects was finally published in January 1998. But after the announcement by the private company and modification of the public effort’s approach, NHGRI staff had to revise the scope of the RFA to accommodate the new strategy and produce a working draft by 2001. Then, the submission dates had to be changed twice for technical reasons, first to June 1998 and then to September of that year. The intricacies of the scale-up process culminated when staff presented the applications to the NACHGR for its second-level review. The Advisory Council spent an unusually long time, eight hours over two days, discussing with staff how to structure, fund, and implement the scale-up program; this was far longer than the Council had ever spent on such a funding plan, before or since. In the end, the NHGRI decided to fund three large-scale sequencing centers to generate the public version of the working draft.

This outcome led to several new administrative changes. First, and most importantly, a different funding mechanism, the cooperative agreement, was used. As discussed below, this change gave NHGRI staff new tools for closer management of the new centers. Second, quantitative goals were made a condition of each award, and use of the cooperative agreement mechanism gave staff more ability to adjust funding levels if those goals were not met or were vastly exceeded. Third, the centers were reviewed yearly by staff and advisors to assess progress toward those goals. Raw sequence data was to be rapidly (every twenty-four hours) deposited in the public databases and only the publicly available data was to be considered in evaluating how well a center had met its goals. Sequence data had to meet specific quality criteria. Investigators were also expected to closely collaborate and coordinate with each other. Finally, staff continued to develop ways to increase the amount of funding available to the centers so that it would be enough to meet the project’s goals. For instance, utilizing the advisory process, larger-than-normal supplements could be provided to those centers that were making outstanding, better-than-expected progress toward meeting goals.

The next few years of the HGP were most demanding for the investigators producing the data, as they had to work under an extremely short deadline and intense public scrutiny, but also for the advisors and NHGRI staff. Weekly conference calls were held with the largest sequencers and their funders. There were also regular, although less frequent, calls with the entire international consortium. Semi-annual face-to-face meetings of the investigators of the largest efforts and staff were held, as was a yearly convening of the entire international consortium. Staff and advisors conducted frequent (at least yearly) site visits to the NHGRI grantees. The sequencing progress reports were submitted biannually and reviewed by staff and the sequencing advisory committee. Staff provided status reports to the Council at each of its three yearly meetings. The director and staff were frequently called on to give progress updates across the government.

Between the funding of these cooperative agreements in 1999 and 2001 (National Human Genome Research Institute) there was much public attention to the HGP, in part because of the perceived competition (significantly media-driven) between the public and private efforts. The intensity of the public production efforts and oversight paid off in 2001 when the international public effort and the private effort simultaneously announced completion of their working drafts. The announcement was made in the United States at the White House by the president and in a coordinated companion announcement by the UK prime minister in London (NHGRI Internal History Archive. Scanned Francis Collins Files, Box 0115, file 012).

While most of the sequencing centers’ efforts during this time were devoted to completing the working draft sequence, they were also able to pay some attention to the development of new technologies to make finishing faster and more efficient. With completion of the working draft sequence, the public sector effort returned its full attention to finishing a high-quality sequence. About 20 percent of the genome had been finished at the time of the working draft announcement; the final 80 percent was finished just two years later in 2003, a full two years earlier than the original goal.

The large-scale sequencing program of the NHGRI was continued beyond 2003 with a series of three-year awards, turning its attention to the sequencing of more human genomes for purposes of discovering the diversity in the human population, of the genomes of many additional organisms, and to medically important sequencing projects. During this time, while sequence production continued to increase, continuing reductions in the cost of sequencing through ongoing technological and strategic improvements allowed the NHGRI staff to shift funds out of the sequencing program and into other programs, reversing the funding increases that had been made at the start of sequencing.