Introduction

1. Origins and Structure of the Volume

The contributions to this volume are varied and unique. Eric Green outlines some of the key legacies of the HGP from the vantage point of the NHGRI and some of the remaining challenges. Opening the “Producing the Genome” section, Elke Jordan, who was the first person hired to help manage the HGP at the NIH, gives a brief account of the early days of the HGP, narrating interesting challenges and surprises. Key to her contribution is how the HGP overcame narrow (even mundane) challenges in its very early years, such as how to hire staff quickly and review grants, which were as important as the more public efforts by Jim Watson to justify the program. Mark Guyer and Jane Peterson joined by Kris Wetterstrand, writing an “anchor” paper for the other staff contributions, provide a deep and challenging analysis of the HGP’s turning points, challenges, and lasting contributions. It is written from the perspective of individuals who were involved directly with most of the key decisions made by the NHGRI during the HGP and other programs, such as ENCODE and the HapMap. Kris Wetterstrand and Jon Lotempio give a detailed account of the HGP’s most celebrated, public, and commented on graphic: the sequencing cost graph.

And Jeffery Schloss delivers a much-needed summary of the importance of sequencing technology development to the overall success of the HGP and genomics efforts, which helps to reframe the HGP as both a scientific and a technological achievement. Elise Feingold recounts central themes in the history of the ENCODE effort and emphasizes the complex scientific legacies of the program, some of which were clearly discernable from the beginning, while others emerged recently, such as the importance of variation to studies of function. Feingold also underscores that the development of a “complete” catalogue of functional elements now seems to be more of a theoretical exercise rather than a goal to be achieved in practice due to the sheer extent of biological complexity uncovered by the ENCODE program. Continuing in a historical fashion, Lisa Brooks discusses how the HapMap and subsequent efforts such as the 1000 Genomes Project learned from earlier inquiries into and the “oversimplification” with respect to human variation research. She argues that the HapMap was defined by pragmatism—cost and existing sample availability plus sequencing and genotyping capacity—and incremental progress, rather than some overarching rationale.

Opening the section “Contextualizing the Genome,” Magdalena Skipper and Chris Gunter address one of the key features of science after the HGP—its influence on publishing. Miguel García-Sancho provides a general sociology of small labs and their contribution to the HGP, with special attention to the situation outside the United States. Detailing the philosophical and conceptual implications of the HGP and genomics, Adam Felsenfeld and Kris Wetterstrand concentrate on the history of the Genome Sequencing Program (GSP) and characterize key philosophical and technical issues. They demonstrate that genomics is primarily a question-generating inquiry, which goes beyond the standard framings of whether genomics operates as a hypothesis-driven or a hypothesis-free science. Joy Boyer and Jean McEwan outline the ELSI Research Program’s history and sound a recurring note from the standpoint of negotiation and strategic compromise.

Christopher Donohue then narrates some of the early challenges of ascertaining polymorphisms and distinguishing errors in sequence data. Ramya Rajagopalan examines key claims of the HGP concerning its future contribution to medicine through a survey of the wide development of clinical genomics efforts by the NHGRI and other funders alongside the profusion of biobanks. She argues that biobanks and clinical genomics implementation have blurred the distinction between biomedical research and clinical care, and that precision medicine aims to diminish the uncertainties surrounding care and diagnosis by discounting the probabilistic nature of genomic medicine, while focusing solely on its purported benefits.

In the final section, “Interpreting the Genome,” Rina Bliss argues that even though human variation programs were designed to avoid issues of racialization and reductionism, they produce them, nonetheless. Eric Villain, Catherine Guaspare, and Michel Dubois emphasize that genetic discussions of behavior and sexual orientation have always been subject to misinterpretation, perhaps even more so in the era of Genome-Wide Association Studies (GWAS), while Alexandra Soulier underscores the many ways in which epigenetics plays a role both in genomic interpretation and in presenting models for sociality and social life. Sahotra Sarkar not only revisits debates over reductionism in the context of genomics, but also illustrates how the unsuccessful reductionist paradigm and failed practices of eugenics motivate public and scientific understandings of the HGP. Sarkar cautions that fantasies of genetic enhancement and the continued desire of scientists to tie complex behavioral traits to genetics remain foundational to genomics due to rhetoric of the HGP—and will remain so as long as practitioners cling to unscientific accounts of genetic causality. Stephan Guttinger and Alan Love reframe the ENCODE debate over genomic function using the exemplar of modENCODE, while Anya Plutynski focuses on how big data approaches in cancer genomics destabilize received accounts of scientific inquiry. Emanuele Ratti and Thomas Stoeger round out the volume by grappling with distinctions about different types of science discussed by Eric Lander and argue that, perhaps counterintuitively, the HGP did indeed aid small lab, discovery science.

We (Christopher Donohue and Alan Love) originally conceived the idea for this volume in the late summer of 2017 after an August meeting at the NHGRI focused on the growth and use of the NHGRI’s History of Genomics Program archive database resources. The NHGRI History of Genomics program was initiated in 2012 by Eric Green, Donohue, and Kris Wetterstrand. Green had been the NHGRI director since 2009 when Francis Collins took over as NIH director. The History of Genomics Program was intentionally established to promote inquiry into the historical, philosophical, sociological, and ethical implications of the HGP and subsequent genomics programs, especially after the “completion” of the HGP in 2003. Although the frame and scope of the meeting were in some senses narrowly technical, we concluded that the archive and History of Genomics Program had become increasingly well known within the scholarly community for its rich primary source material through several small conferences and publications (e.g., Green and Donohue 2018). This, in conjunction with the thirty-year anniversary of the commencement of the HGP by the NIH (in October 1990), provided a timely and appropriate rationale for a workshop that was organized around the theme of the HGP and genomics. Furthermore, despite the substantial secondary literature on various aspects of the HGP and other genomic inquiry, the editors discerned that there was more than a need to fill specific gaps in existing literature—there also was potential for a novel and significant intervention into the scholarship on genomics that might both upset some comfortable narratives and set an agenda for future investigations by science scholars interested in the sprawling space of genomic science.

With these factors in mind, the History of Genomics Program convened a conference in November 2018 that sought to address the HGP and genomics from a variety of perspectives, including historians, philosophers, scientists, and sociologists, while also covering the full range of topics in genome biology and genomic medicine, with social and ethical implications strongly in view. Because of the conference’s location on the NIH campus, a number of program staff (including some who had recently retired from federal service) from the NHGRI who played critical roles in the administration and guidance of the HGP (and subsequent genomics programs) were asked to give papers detailing their specific experiences, vantage points, and expertise. Topics included the GSP, human variation programs funded by the NHGRI (e.g., the HapMap and the 1000 Genomes Project aimed at comprehensively describing all common human genetic variation), genome sequencing technology development (e.g., the $1000 Genome), ENCODE, and the ELSI research program. This fusion of both “outsider” scholarly perspectives and “insider” scientific program perspectives made for a unique workshop and now is augmented and enhanced by the juxtaposition of refined and developed contributions to the present volume. Inspiration for this complementary combination of insider privileged access and outsider critical distance borrows from Robert Merton’s dated, though still useful, analysis (Merton 1972).

Although there have been some changes in contributors and topics since the 2018 conference, the fusion of diverse perspectives remains integral to the volume. However, reflections on the scope and content of the project among the editors and meeting participants, including the scientific staff, signaled a need to circumscribe the reach of the volume, keeping it focused primarily on the roles of the NIH and the NHGRI. This was because most contributions from humanist scholars and program staff focused on these roles, in part because they reflected their primary domains of expertise. Thus, while this volume provides unique insights into the development of the HGP and genomics from the late 1980s to the present, it does so primarily (but not exclusively) from a United States, NIH-centric perspective. This does not mean that the contributions of other US government agencies, such as the Department of Energy (DOE), or those of other national sequencing projects, such as occurred in the United Kingdom, France, Germany, Japan, and China, were unimportant. It also does not imply that various US sequence and mapping centers involved in the HGP and other genomic initiatives, such as Washington University in St. Louis, the Broad Institute (MIT/Harvard), Baylor University, and others, were not key players on the landscape of genomics. In fact, by restricting the scope and content of the present volume primarily to the US NIH/NHGRI efforts, we argue for the opposite: the importance of these distinct extramural project locations, as well as diverse governmental and national efforts, necessitate separate, specific investigative analysis. Indeed, laudable efforts to internationalize the HGP and to underscore the foundational importance of model organisms sequenced comparatively alongside the human genome have been undertaken (e.g., García-Sancho and Lowe 2023).

Even with this limitation in scope, the present volume nonetheless makes a significant and novel methodological contribution to the scholarly analysis of genomics. Although another series of papers comprising a special issue on “Genomics and the Human Genome Project” for the Journal of the History of Biology was the first to make use of the NHGRI’s extensive archival holdings (Green and Donohue 2018), this volume continues and expands the utilization of the NHGRI archive in a significant number of its papers. This holds not only for historical papers (where this might seem most natural) but also for chapters focused on conceptual questions or sociological issues, including those from the scientific staff. Overall, Perspectives on the Human Genome Project and Genomics seeks to bridge many areas of critical scholarship that have hitherto been separate, often because of disciplinary differences, through an alloy of heterogeneous accounts of the HGP and genomics. This welding together of historical, philosophical, scientific, and sociological perspectives on topics in genomic science and medicine, as well as some interventions by authors into the history of technology, provides a tactical pivot point for future scholarship undertaken in the fourth decade of genomics research.

2. Opening the Archive and Inviting a Plurality of (Often Critical) Perspectives

Prior to the first collection of archivally-informed papers (Green and Donohue 2018), scholarship on the HGP and other genomics programs was based almost entirely on published papers, printed news sources, and selective interviews with high-profile scientific personalities. Although Robert (Bob) Cook-Deegan’s excellent work on the early years of the HGP balanced different perspectives, including his role as both an insider and an external critic, subsequent scholarship on the large and heterogeneous field of genomics has focused almost exclusively on “controversies,” such as “the race to the finish” between the public program and Celera (see, e.g., Davies 2002; Shreeve 2007; McElheny 2012). This is understandable given that such episodes are, by definition, the most visible. Nonetheless, while the present historiography incisively highlights many issues of importance, it also obscures, often by constructing seeming continuities and purportedly shared worldviews. From the standpoint of archival evidence, discontinuities and individual agency are more visible. This hindsight from archival sources makes it possible to summarize the dominant virtues of existing historiographic discussions, while also offering an opportunity to criticize aspects of a pervasive consensus around postgenomics.

Historically, there has been a significant disquiet among humanists with scientists writing critical commentary in the history of science or philosophy of science (Brush 1995; Falk and Falk 2007; for a classical formulation of the problem, see Williams 1975; Hull 1979). Although this general disquiet has lessened, scholars working in history and philosophy of biology have cultured and valued a continual exchange with biologists and geneticists despite these concerns. Although theories and politics have often been hotly debated, the status of geneticists vis-à-vis historical and philosophical studies of genetic science has not been seriously questioned, in part because prominent geneticists and molecular biologists have written useful (if sometimes somewhat narrow) histories of their own discipline (e.g., Morgan 1932; Sturtevant 2001; see also Lewontin 1976). Indeed, within history and philosophy of biology as an area of inquiry (going back to the early 1970s; e.g., Grene 1974), philosophers and life science researchers have had many fruitful interactions (e.g., Falk 1986; Gilbert and Sarkar 2000; Love 2015).

As noted, many discussions of the HGP and genomics have revolved around its social and ethical implications, and especially some of the more extravagant and controversial claims made upon its commencement, including open questions about whether the HGP would herald a new era in eugenics (Sarkar and Tauber 1991; Tauber and Sarkar 1992). Philip Kitcher reflected a shared sentiment when he said, “We don’t need the HGP to inform us that there are important social problems that face our society” (Kitcher 2003, 280), but noted that critics of the HGP had overestimated the extent to which the project would lead to a resumption of eugenics or other forms of ideological reductionism (though such discussions continue to be relevant; Sarkar 2021). Nobel laureate Jim Watson was acutely aware that the availability of the DNA sequence of the human genome would not only yield significant scientific dividends but also lead to fraught social and ethical questions, including questions of reductionism and eugenics. It was for this reason that Watson proposed the creation of the ELSI Research Program in 1990 to help guide and foster objective research into the societal implications of genomic science. Such research—and the NHGRI’s continued support of it—has proven more than necessary from the onset (though perhaps inadequate) because there have been continual attempts to deploy genetics and genomics data for problematic endeavors, the description and implications of which have constituted a large portion of the scholarship on genomics and the HGP since its inception (chapter 10, this volume). And in this regard, Watson’s decision also embodied a certain irony, given his long history of uttering eugenic statements, as well as his avowal of racist views on intelligence.

The ELSI Research Program—a full history of which needs to be undertaken by some scholar or a group of scholars in the future—remains a unique undertaking in both the science and ethics realms. It is unique insofar as it is the largest, continuous funder of studies into the ethical, legal, and social implications of genomics research, having its origins in the very beginnings of the HGP itself. The history of ELSI, amply attested to by the NHGRI archives, is in many ways a reflection of the development of genomic science. Along with the mapping, sequencing, and mammalian genetics branches, ELSI was one of the four original branches of the NCHGR (Meslin et al. 1997). Early work supported by the ELSI research program developed around the ethics of genetic counseling, particularly cystic fibrosis testing, and the uses of genetic information. Because work on the ethics of cystic fibrosis screening began in 1989 with the efforts of Elizabeth Thomson at the National Institute of Child Health and Development, ELSI inquiries technically predate the HGP itself (NHGRI Internal History Archive. Scanned Files btw NHGRI & Outside Orgs 1984–1988/15 Ad Hoc Cmte on CF Carrier Screening 1991–1992, folder 002).

In her 1991 remarks to Congress, Bernadine Healy noted that one of the crucial unknowns of the HGP was the potential for misuses of genetic information; she detailed a suite of projects around protection of privacy and other social implications of genomic science. Healy highlighted that there were, “projects aimed at developing professional guidelines for the confidentiality of genetic records, clarifying the legal foundations of genetic privacy, assessing the social implications of genetic identification techniques, surveying collections of genetic data, and evaluating the potential uses of genetic information by insurers and employers” (NHGRI Internal History Archive. Scanned Francis Collins Box 1373–2, folder 029). Such efforts arguably paved the way for the Genetic Information Nondiscrimination Act (GINA) of 2008, which prohibits employment discrimination based on genetic information (EEOC statutes). The ELSI Research Program also served as a forum for discussions surrounding Charles Murray and Richard Herrnstein’s 1994 book The Bell Curve (NHGRI Internal History Archive. Scanned Box 1307, folder 017). Early ELSI funding—all of which was investigator-initiated—also supported projects around the sociology of genomics, intellectual property questions and commercialization, as well as human subjects’ issues. Early ELSI grantees also pursued projects relating to genetics, scientific racism, and the role of the HGP and genomics in stigmatization. (One of the contributors to this volume, Sahotra Sarkar, was an early ELSI grantee on genetics and reductionism.) Historical monographs on perceptions of cancer risk as well as histories of genetic testing continued to be emphasized in ELSI funding through the early 2000s, as evidenced by detailed minutes and materials from the “NHGRI Planning Workshop Bioethics and Humanities Research: Genetics and Worldviews” held in 2002 (NHGRI Internal History Archive. Scanned Box 035, folder 017). As outlined by Boyer and McEwan, the community consultation process was an outgrowth of the ELSI programs priorities and a natural extension of the ELSI’s research programs historical interest in issues of race, reductionism, and informed consent. Discussions of ethical considerations and how these should be informed by community input began very early in the development of the HapMap (see e.g., NHGRI Internal History Archive. Scanned Mark Guyer Box 13/1, HapMap folder).

Normative discussions about how to integrate the science and ethics of the HapMap are amply documented in the NHGRI history archives. However, the reasons for misinterpretations and misunderstandings surrounding the role of the ELSI research program are various. In the context of the HapMap, ethics concerning community consultation, proper use of samples, and proper uses of the data generated were such an integrated part of the program that the ELSI efforts were perceived as less than prominent than they were in reality. Arguably, this was a consequence of how the ethical stance became normalized inside of the HGP itself. Because the ethics of genomics are sometimes identical with how the science is designed and motivated, many of the stated rationales can be subject to misinterpretation (see NHGRI Internal History Archive. Scanned Francis Collins 7108, folder 004).

One charged dimension of this commentary is that the HGP, as well as other genomics initiatives, reintroduce (either intentionally or not) reductionistic biological accounts of race into medicine and the wider public imagination (e.g., Reardon 2009; McMahon 2019). Accounts like these are useful and quite incisive (especially Reardon 2009), but they often partially obscure other factors. The HapMap does indeed at times somewhat rhetorically mirror the Enlightenment, Romantic, and nineteenth-century continental grouping of races. This was less a product of an overall plan to do so than a combination of converging factors: (1) various sequencing and genotyping facilities had capacity, money, and the strong inclination to join an international project (Beijing Genomics Institute [BGI] and Riken); (2) an international project was desirable after the conclusion of the “initial sequencing” of the human genome; (3) the over-general but still useful assumption that common variation drove common disease phenotypes (called the common disease, common variant hypothesis, or CDCV) and that GWAS studies would be easy to properly power with a well-marked map from a few geographically distributed populations; (4) it was possible to derive a considerable amount of clinically and biologically significant variation from a relatively small number of samples; and (5) preexisting community consultation expertise on the group within the Yoruba in Nigeria. The NIH program officers associated with this effort did not promote any type of racist or “exclusionist” ideology associated with race science and eugenics (Donohue 2020; Turda 2020).3 And, just as importantly, the HapMap was designed to only interrogate differences in a limited manner in haplotype structure between and across populations, rather than sample human genetic diversity on a global scale, the latter being the goal of the Human Genome Diversity Project (HGDP) (see Clark et al. 2003). However, there has been a valid recurring concern that genomic medicine, and more recently personalized medicine, associates specific variants or genes with specific groups and also associates specific groups with specific disease phenotypes (Roberts 2011), which seemingly recapitulates racist and eugenicist arguments (Comfort 2012). These concerns were exacerbated by historical controversies, such as the identification of BRCA1 and BRCA2 as “Jewish genes” in the mid-1990s in conjunction with their observed higher incidence in individuals of Ashkenazim ancestry (Mozersky 2013), or the more recent controversy over whether specific medications for heart disease and hypertension could be aimed at pharmaco-genomic targets that some argued are more predominant in African American patients (see Kahn 2013). Such reasoning, especially if generalized, might lead to the development of a specific class of therapeutics in the pursuit of “racialized medicine” for “black patients” and “black diseases,” though there are compelling counterarguments against such reasoning in the context of precision medicine (Bonham et al. 2016; Interview with Charles Rotimi 2015). As important is the sustained discussion of the risks of racialization and reductionism, and decades of overhyping about whether precision or personalized medicine is worth pursuing, especially given scarce and unequal resources for community health interventions (Tabery 2023)

In the specific context of genomics programs investigating human variation, such as early efforts in the late 1990s to ascertain patterns of polymorphism in large-scale sequence data, there has been a complex discussion of whether, and if so how, NIH-funded studies of variation (e.g., the HapMap) recapitulate—purposefully or not—traditional “folk” racial categories. In the attempt to capture the full spectrum of rare and common human genomic variation for the purposes of developing increasingly robust GWAS, many have been concerned that these programs reinforce these problematic categories. A variety of historians, sociologists, philosophers, and science, technology, and society scholars have been involved and even stimulated this debate. Most notably, the philosopher Quayshawn Spencer argues that clustering algorithms in software programs such as STRUCTURE, which were developed to handle data emerging from the HGDP, do sort populations into continental groups based on similarity in population structure, where allele frequencies are used to determine whether an individual may be assigned (based on genotyping data) to a specific subpopulation, while those same subpopulations are clustered using allele frequency differences across and between populations. According to Spencer, these groups are precisely those of the folk racial categories outlined by Johann Friedrich Blumenbach more than two hundred years ago. Thus, at least for some, a reductionism of this type within human variation programs is unproblematic because folk races are indeed real and justified by genomic data (Glasgow et al. 2019). While provocative and challenging, Spencer’s discussion fails to bridge the divide between what Blumenbach himself meant by his categories and those that are defined by modern clustering algorithms. No population geneticist today would subscribe to either Blumenbach’s vitalism (which was part of his critique of preformationism) or his account of natural history and causation out of which his discussion of “varieties” sprang (see Eigen and Larrimore 2012; Rupke and Lauer 2018). Thus, while both efforts do categorize, the meanings of those categories or groups are so distinct as to make comparison difficult, in part due to reliance on different kinds of metaphysics and the types of evidence utilized for their construction.

A central element of these discussions is whether the NHGRI, through the development of efforts such as the SNP Consortium, the International Haplotype Map Project, and the 1000 Genomes Project, sought to construct a type of “race map.” The concern was that this type of “race map” would involve the decision to include specific populations primarily on the assumption that those populations possess specific or unique genomic attributes that are not present in other populations. Even if done ostensibly for epistemic reasons, many scholars argue that genomic human variation programs have unwittingly reconstructed racial groups and justified them via a form of genetic reductionism. “If the late 20th century was characterized by the deconstruction of race, then it is being reconstructed around a biomedical and informational paradigm at the beginning of the 21st century” (Wailoo et al. 2012, 82). According to many, NHGRI-funded studies of human genomic variation consciously “make” groups, thereby reducing social categories and social complexities to genetic variants and genes (e.g., Reardon 2007; Gannett 2014). The resulting conferral of genetic group identity is not mitigated by increasing participation so that genomic studies exhibit a more diverse demography, nor will a particular framework of liberal governance eliminate the resulting problems. Contributions to the present volume engage this complex discussion willingly and devote a great deal of attention to the complicated discourse around human variation research, including ethical, philosophical, sociological, technological, and historical perspectives (see chapters 5, 10, 11, 13, and 16).

3. We Have Never Been Postgenomic: Beyond Standard Assumptions

Although various aspects of genomics and the HGP have been well represented and have generated an interesting and contentious scholarship throughout its thirty-year history, we find one recurring idea across these analyses misleading and problematic: “postgenomic.” Scholars have defined the postgenomic as an episteme or “a condition” (Müller-Wille and Rheinberger 2012; Rheinberger and Müller-Wille 2017), though sometimes with a tacit commitment to its origination in the wake of the HGP. For example, Reardon underscores that the “postgenomic condition” is “this turn to the question of meaning—the question of the uses, significance, and value of the human genome sequence” (Reardon 2017, 2). Reardon also laments that, after the “completion” of the sequence (assumedly in 2003), there has been a marked decline of interest in genomics on the part of the scholarly community; unlike the drama displayed in the generation of the sequence itself, now “there are no clear heroes and villains” (Reardon 2017, 2). For Richardson and Stevens, postgenomic refers to “a break from the gene-centrism and genomic reductionism of the genomic age” (2015). However, this purported break is difficult to detect given the controversies noted in the previous section that are specifically about genomic reductionism in relation to categories of race. The presumption seems to be that scientists were overly naïve during the HGP and have woken up to a more variegated biological world in the wake of its completion. “Scientists . . . trace a path from a simplistic, deterministic, and atomistic understanding of the relationship between genes and human characteristics, towards, in the postgenomic era, an emphasis on complexity, indeterminacy, and gene environment interactions” (Richardson and Stevens 2015, 3). Most of the contributions to this volume challenge this framing and periodization, whether in terms of what dominant paradigm is presumed for those who were involved in the HGP, when the HGP “finished,” or in terms of how “they” (aggregated as a single, monolithic entity) have responded to unexpected discoveries.

This understanding of a postgenomic state often assumed that the HGP has been “completed,” an assumption abetted by the public rhetoric of a White House celebration in 2000 when a draft sequence was achieved or later with the end of formal funding of the HGP. This visibly marked “closure” masked the complex, evolving organization of projects that had multiple timelines, periodizations, and activities. Although the generation of the 2000 and 2003 human genome datasets represented an exceptional achievement and important advance in the knowledge of the field, the HGP was not actually finished in several critical senses for almost another twenty years (Nurk et al. 2022).

The standards of completion have changed over time and vary across genomic sciences (discussed in more detail below). However, even with the most up-to-date build of the reference sequence according to the latest evaluative standards, gaps and unresolved regions of biological significance remain (interview with Deanna Church in 2019). With respect to these precise technical standards, it is not clear the HGP can ever be considered “finished,” in part because the human genome reference is being continually updated (International Human Genome Sequencing Consortium 2004; for discussion of issues associated with the reference genome, see Ballouz et al. 2019; for specific difficulties associated with structural variants, see Church 2019). NHGRI leaders have to a certain extent “owned up” to this uncertainty (Green 2020). Nevertheless, the recently completed efforts by the Telomere-to-Telomere Consortium (T2T) emphasize only “a first complete assembly of a human genome”4 (see Nurk et al. 2022). Significantly, the sequence itself was derived from a “a complete hydatidiform mole” with a 46, XX karyotype, which was chosen to artificially limit heterozygosity and create an artificial system ideal for a nearly errorless, contiguous, complete sequence assembly. Such a system would never have given rise to a living organism, much less a person, and is therefore somewhat misleadingly described as a “human genome.” It can be more accurately described as a gapless, complete sequence from a human cell line or human DNA.

Setting aside these technical issues related to the meaning of “completion,” the framing of a singular “postgenomic” condition or era assumed by some scholars can be seen as well motivated from the point of view of social critique. Public pronouncements by scientific leaders, the president of the United States, and the popular press are relevant to analyses concentrating on the social impact of the HGP. Even though this orientation might align poorly with the perspective of scientists who do sequencing work or government officials who organize and fund it, we can comprehend why this orientation was adopted. Put bluntly, whose perspective matters for deciding when the HGP is genuinely “finished”? However, the present volume encourages a pluralization of perspectives that challenge the claim that there was a single “postgenomic” fault line that serves as a boundary line for the “postgenomic era.” The periodization associated with public pronouncements cannot remain the dominant narrative considering the multiple timelines and activities exposed by both technical considerations and a richer grasp of the historical heterogeneity contained within the label “HGP.”

However, it is not just the periodization that is problematic; continued associations of the HGP with a deterministic and reductionist research outlook and postgenomics as an era of indeterminacy and complexity miss the heterogeneous combination of contingency and innovation present throughout genomic initiatives. The HGP harbored inherent indeterminacy because even though the goal was conceived—transforming biological sequence into computationally analyzable data—the knowledge gains from achieving it were largely unconceived. The HapMap was in a sense deterministic because the grantee community assumed that between-group haplotype variation was sufficiently important for studies of human disease to study multiple, ancestrally diverse groups. Some contributors to this volume provide examples that illustrate how associating the HGP with determinacy and reductionism is problematic, even aside from the uncertain meaning of those terms in the context of genomics. While some of the rhetoric surrounding the HGP could be construed in such a way, the consensus among the grantee community and program staff was that the HGP represented both a limited dataset and the tremendously important beginnings of a community data resource, which could be built on in a variety of ways. The divergence between this consensus and perceptions among the public and other stakeholders constitutes an important fault line. As Mark Guyer, Jane Peterson, Kris Wetterstrand, and Elke Jordan all underscore in their contributions, the HGP was intended less as a view of human nature than a series of management efforts for scientific activities intended to overcome significant issues and problems pertaining to producing a limited dataset under specific technological conditions with limited time and funding. However, this perspective alone would have been inadequate to convince Congress to back the multibillion-dollar budget required.

Circumscription of the dataset was a necessity if the data resource was to exhibit the requisite quality. Early physical and genetic maps harbored significant issues in this regard. Sequencing through the late 1990s was not only costly but remained error-prone. Prior to 1997, there was not a universally robust, reproduceable, and accepted statistical measure nor a sufficient criterion for quality or quantity of sequence to consistently determine the difference between a DNA sequence polymorphism and an error. As Donohue outlines, the distinction between a polymorphism and an error emerged from a series of informal estimations and provisional understandings that were historically contingent as well as dependent on what sequencing and analysis could achieve cost-effectively.

Additionally, many technologies were not scalable. Adam Felsenfeld and Kris Wetterstrand’s contribution argues that the HGP and subsequent genomic programs use limited and incomplete but illustrative datasets as “scientific question generating devices.” Genomic initiatives, from this perspective, are primarily about facilitating new biological inquiry rather than solving existing problems, though again this was not how they were sold to broader audiences or why they are funded, which focused instead on the contributions of genomics to the field of medicine. They remind us that large, coordinated science projects, such as the HGP, HapMap, or ENCODE, are difficult and unwieldy to coordinate given the number of participants involved. Also, they are ruthlessly tied to cost points, which means that project development and its goals are often a series of negotiations between the amount of funds available and the scientific achievement that can be secured with the amount of sequence and analysis that (it is assumed) can be generated with those funds. The data generated often result in different questions being addressed, either out of necessity or due to novel discovery, and in known questions being addressed less than satisfactorily. In fact, the latter situation accounts for large portions of the controversies that cling to genomics initiatives.

Keeping in mind this fragile balancing of competing factors, the contingency of the HGP (in particular) and other large-scale genomics programs increases in salience. Much of the history of the HGP and similar initiatives in genomics indicates that they need not have succeeded, despite being punctuated with clear successes. Especially in the early years, these efforts are best characterized in terms of fragility and contingency with progress and local breakthroughs along the way. This characterization arises not only from the juggling involved in securing adequate funding but also by the sustained differences of opinion obtained among associated program teams. The HGP and subsequent projects were animated by many intense discussions and internal controversies in which individuals occasionally vehemently disagreed about what the HGP goals were and, more frequently, whether its methods could genuinely achieve them.

In addition to the difficulties with characterizing the HGP as a monolithic project defined by determinism and reductionism, the standard periodization also needs to be criticized for presuming that subsequent (postgenomic) efforts in genomics are somehow distinctly concerned with and marked by indeterminacy and complexity. Genomics programs such as HapMap and 1000 Genomes, or various uses of GWAS data and applications of genomic medicine, are sometimes distinctly not marked by complexity but instead might be best described as “reductive.” Postgenomics, even if treated as a distinct era, is not necessarily immune to standard criticisms or obviously unproblematic and an improvement over the HGP. The contributions to this volume limn the contours of the complexity, indeterminateness, and sometimes lack thereof for both the HGP and other genomics programs. Several consistent themes emerge across the contributions, including distinctive and unexpected mixtures of robustness and contingency, as well as challenges and modifications to our understanding of “big science” or “collaborative science” in the context of the HGP and genomics, which seriously interrogates the traditional dichotomy between “investigator-initiated projects” (so-called “small science”) and “center-directed” efforts (so-called “big science” run centrally out of a government agency).

Eric Lander and others have been keen to point out that much of genomics is “hypothesis limited science” where investigators approach genomic data with a particular set of prerequisites, hunches, or inclinations. However, due to the availability of genome-wide data, they often discover elements that they were not initially looking for and sometimes leave behind the initial questions that motivated attaining and examining the genomic data. Lander, echoing a sentiment often expressed by others involved in genomics programs, observes that when people work with genomic data, they have the perception that they harbor no other presumption than what they are looking for is “genetics.” This does not mean that the investigative strategy is unproblematic. For example, looking for genetics is sometimes a problematic approach and encourages neglecting relevant causal variables, such as those related to the environment. However, it does mean that centrally controlled, “cabalistic” portrayals of genomic juggernauts are likely wide of their mark. Characterizations that keep the HGP tools and genomics techniques foregrounded in the development of genomic science are more apropos. “The ability to take comprehensive views has pretty much burned down . . . the cartoon version of a hypothesis-driven experiment, where you are only allowed to go in with one idea. . . . Now you go in and you look at what’s going on” (Interview with Eric Lander 2019). Lander’s rejection of a cartoonish version of hypothesis testing points not only to an interpretation of genomics and the HGP that is highly pragmatic, but also to an account where the project details and scientific consequences are decidedly more flexible than what is presented in some accounts. At the same time, it also points to genuine blind spots in genomics programs, such as searching primarily through genomic data (as opposed to, until relatively recently, exposome or other data that might inform our understanding of health and disease) because that is the lamppost these scientists work underneath.

One specific example that demonstrates this flexibility in execution, while simultaneously emphasizing the difficulty of issues encountered within the HGP and solved through a process of negotiation and consultation among grantees and scientific staff, was how various actors navigated the bacterial artificial chromosome (BAC) library dilemma. During the fall and summer of 1996, multiple issues with the BAC libraries used for sequencing were identified by program staff due to a series of revelations by the grantee community. First, many of the libraries in use at the time were not properly consented or their consents were nonexistent, necessitating the funding of newly developed, properly consented BAC libraries, which also required updated research participant guidance by the NIH and the DOE. Second, program staff and grantees realized that if a single donor was used for a substantial portion of the BACs necessary for sequencing the HGP, then it would be possible (though not likely) for this individual to be identified through some means in the future. Although investigators and program staff understood that the initial sequence, though significant in and of itself, would be supplemented with substantial downstream analysis, NHGRI program staff predicted that this supplemental analysis would most likely focus on disease genes and other disease architecture. This meant that there was a genuine worry that this individual would read about some diagnosis or risk assessment in the press and infer a connection to themselves. Thus, it was imperative that for this and other reasons the HGP move to a “mosaic” approach of multiple donors for the newly developed BAC libraries. The fact that a single library was used for most of the sequence, while still composed of a mosaic of individuals, following the letter, but not necessarily the spirit of this guidance, continues to provoke discussion, with a recent argument that the library discussions took the form that they did in response to the ethics of the time, which would not necessarily be followed in the same way today (LoTempio et al. 2025).

In the context of directing a successful large sequencing pilot with multiple centers, during which time there were the first indications of the successes and challenges of large-scale sequencing, Mark Guyer and Elizabeth Thompson of the ELSI Research Program and others had to immediately shepherd through not only a new consent process, but to implement the development of new BAC production facilities. This process, which was relatively unknown outside of the grantee and scientific program community until 2024, illustrates several attributes about the HGP as a project that are described in various chapters of this volume. Perhaps most importantly, the HGP was not directed by the NHGRI; rather, the NHGRI and grantee community worked together, sometimes harmoniously but sometimes with significant discord, to solve specific, difficult problems related to the goals of producing efficient and high-quality sequence in an ethical manner.

Although this coordination produced disagreement, the project nonetheless proceeded. Even during the “friendly fire” incident regarding sequencing strategy—a controversy that was very heated for a relatively short time—not one of the major sequencing groups (later called the “G5”) left the HGP. However, just as importantly, the decision-making process around BAC library mosaicism serves to illustrate how closely technology, biology, and ethics were intertwined, to such an extent that the development of new libraries forced a pause in the project and necessitated a far-reaching discussion that lasted the better part of a year in the mid- to late 1990s, at the height of the HGP, which saw the transition from mapping to sequencing. While that process still strikes some as a less than ideal, it nevertheless shows the centrality of ethics to the HGP, regardless of the divergent views about those ethics.

4. We Have Always Been Postgenomic

The contributions to the present volume demonstrate that genomics, whether in its ambitions that began under the aegis of the HGP or in its continuation in one or more subsequent initiatives, has always been marked by: (1) a heterogeneous (at times motley) community of inquiry rather than monolithic centralization; (2) incremental results, datasets, and tools more so than overarching theoretical frameworks or approaches; and (3) question generation and sometimes even limitations on possible questions rather than universal biological generalizations. Nonetheless, we are aware that it is precisely because there is a heterogeneous community with incremental results and new tools that a few researchers will sometimes advocate overarching theoretical frameworks and situate their efforts in terms of a purported episteme. At the root, our plea is for the studious avoidance of simple, static classifications and more attention to ubiquitous nuance and polysemy.

One of the most significant and underanalyzed distinctions between the HGP and subsequent genomics programs lies in the distinction between “complete” and “comprehensive.” With the completion of a “first complete assembly of a human genome,” this distinction has been (perhaps accidently) thrust back into the spotlight (see Miga et al. 2020 and Nurk et al. 2022). Completeness was one of the hallmarks of the HGP, and it retains a variety of meanings. Rather than being dogmatic, researchers are often heuristic in their reasoning and pragmatically (hyper)flexible (even to a fault), at times appearing nonchalant. Genomics projects after the 2000 draft sequence often used the term “comprehensive” to describe their goals (see for example Elise Feingold’s contribution on ENCODE) rather than completeness. The subtle shift from complete to comprehensive as project aims not only makes any account of the postgenomic more difficult, in part because there is no longer a bright line of demarcation for “before” and “after” based on the completeness goal, but it also signals how the research community itself grew in awareness of what kinds of epistemic targets should be in the investigative sights of genomics. These targets consisted in iterative endeavors that involve layering different results, datasets, and tools from genomics that, in the process, open new horizons not previously recognizable. Put differently, the targets could not be predefined because the methodologies creatively yielded something unanticipated, such as general principles of metazoan genome operation (chapter 17, this volume).

Completeness was clearly in view from an early stage even if not explicitly articulated. Throughout the HGP, both the research community and the NHGRI emphasized the goal of the project being a “complete human genome sequence.” This appeared in various strategic plans, in the public contest with Celera, and in multiple press releases and public discussions leading up to the “Initial sequencing and analysis of the human genome” published in Nature in 2001. Indeed, in this critical paper, the word “complete” is used sixty-one times. Importantly, “complete” and “finished” are used synonymously. “Completeness” and “finished” are both social and technological measures that are commonly agreed upon, though not always explicitly articulated; few investigators set out to achieve incompleteness or to not finish despite there always being more research to undertake. (Although perhaps conceptually apt, describing the HGP and other initiatives as the never-ending natural history of biological molecules would likely be met with blank stares.) In the late 1990s, the International Human Genome Consortium (IHGC), which included the twenty or so major partners in the effort to sequence the human genome, agreed that having a “complete” sequence meant resolving the vast majority of euchromatic regions of the genome, but leaving aside the heterochromatic regions because they are (and continue to be) very difficult to sequence and analyze.

The NHGRI and IHGC agreed on a qualified definition of completeness that privileged various assumptions about sequence quality and contiguity, as well as which regions of the genome would be most important to resolve. This agreement relied on some background assumptions about function and variation held by members of the NHGRI and the grantee community (e.g., the functional significance of euchromatic genomic regions and the relative unimportance of heterochromatic regions). However, more significantly, discussions of completeness were motivated by sophisticated accounts of how much biological information could be extracted from the DNA sequence given the material costs of sequencing and analysis. Thus, in 2000, the genomics community agreed that trying to resolve all gaps and errors in the genomic sequence, and taking on the costs associated with that effort, would largely reduce the value for most of the sequence that could be relatively simply resolved (compared to other regions), and then released to the wider community for research benefit. By 2004, when the follow-up paper “Finishing the Euchromatic Sequence of the Human Genome” was published in Nature, signaling the technical “end” of the HGP, the research community had to address the reality that even the euchromatic portion of the human genome sequence still had a few hundred gaps. Until 2022, with the work of Adam Phillippy’s group in concert with a large genome research consortium, “asymptotic” completeness (or absolute sequence completeness) had been deemed of secondary importance (conversations with Deanna Church and Adam Felsenfeld 2018 and 2019).

What this illustrates is that even though the HGP placed an emphasis on completeness, its meaning was dynamic and revised in light of practical challenges and epistemic interests, and even on occasion dismissed altogether. Even more significantly, subsequent genomics programs do not have completeness as an explicit aim. For example, the HapMap Phase 1 Project (c. 2002–2005) also put an emphasis on resolving all common haplotypes in NHGRI internal program-related correspondence, but in the public paper communicating the results, the emphasis shifted from completeness to “comprehensiveness.”

A comprehensive search for genetic influences on disease would involve examining all genetic differences in a large number of affected individuals and controls. It may eventually become possible to accomplish this by complete genome resequencing. In the meantime, it is increasingly practical to systematically test common genetic variants for their role in disease; such variants explain much of the genetic diversity in our species, a consequence of the historically small size and shared ancestry of the human population (International HapMap Consortium 2005, italics added).

The change in vocabulary from “complete” to “comprehensive” in many programs was emblematic of a more nuanced understanding of genomics as a science, which began to consistently underscore the partial, though informative nature of datasets in a world of multiple genomes. The NHGRI and the research community began to realize that genomics, rather than furnishing a complete product, would be better conceptualized as an inquiry engaged in the production of finite, imperfect datasets that would allow for the testing of certain assumptions and the development of additional conjectures. While it was indeed possible to sequence a “whole” individual genome, subsequent genomic analyses of variation or functional elements demonstrated that completeness in this sense was an inappropriate standard because intermediate quality data from multiple sequences could be aggregated to much the same effect through bioinformatics tools. Although in the beginning of single nucleotide polymorphism (SNP) discovery efforts there was some discussion about how the 5 kilobase (kb) map would lead to a better understanding of the genetic contribution to common disease, it was quickly realized that more refined iterations of the HapMap would be needed, both with better variant ascertainment techniques and more populations.

Overall, the variance between the two terms—completeness and comprehensiveness—is as much a paradigmatic shift as an implicit acknowledgment of the research community’s and NHGRI program officers’ many goals with sequence data, underscoring the multitude of projects and the profusion of scientific aims. The use of these two terms is part of the recognition that while many practices have transferred from the HGP to other genomics initiatives (e.g., consortiums, rapid release of data, data quality control and maintenance, pooling of methods, and genome-wide applications of standard methods), the goals of genomics have both changed and proliferated, from a seemingly primary standard subject to shifts in meaning (e.g., completeness) to a variety of flexible, project-specific criteria (e.g., comprehensiveness with respect to functional elements or variation). This pattern is highly visible in publications from different projects, such as for the 1000 Genomes Project: “[it] set out to provide a comprehensive description of common human genetic variation by applying whole-genome sequencing to a diverse set of individuals from multiple populations” (The 1000 Genomes Project Consortium 2015, 68).

What “comprehensive description” means in genomic science is open to interpretation. In the context of the 1000 Genomes Project, comprehensive pertains to common and rare variation and is calibrated with respect to a finite amount of analyzable sequence and the maximum amount of information and analysis that can be extracted from the available amount of sequence for a particular cost point and for a particular purpose: the understanding of the genetic contributions to health and disease. This does not mean that further analysis of more sequence at a greater depth and lower cost would yield the same conclusions. In fact, the history of variation research has seen several hypotheses thoroughly reexamined on exactly this basis. Comprehensiveness, according to most investigators, simply refers to the limits of testing with a given dataset or the idea that “we’ve tested everything we wanted to test” (Conversation with Adam Felsenfeld 2019). Clearly, this can be easily misinterpreted when appearing in broader contexts of communication. The reality is that what researchers want to test evolves, which means that project goals related to comprehensiveness are subject to conceptual change over time.

5. Parting (Introductory) Thoughts

The contributions of this volume adopt a plurality of perspectives that yield a heterogeneity of views. Together they offer both a timely and unique collection of scholarship that emphasizes the NIH role and perspective on the HGP and genomics. While not disparaging prior approaches that have concentrated on a strict periodization or displayed the utility of critical commentary, we think that the enormous complexity of genomics and the HGP necessitates a more nuanced picture of a still-evolving science that resists easy characterization and categorization from any single perspective. Thus, even within its delimited frame of reference and not discussing in detail other national and governmental partners, this assemblage of interdisciplinary scholarship tries to move beyond overly simplified conflict accounts, as well as the triumphalist narratives that present the HGP as an inevitable revolution (among other shibboleths). Insights can be gained from such perspectives, but with the ever-increasing availability of archival data and the widening of the scope of investigative work on the HGP and genomics, it is time to appreciate that scholars and scientists can both work separately (and together) to build new and more nuanced accounts. The value is double-edged because this volume serves as both a template of collaboration and as a challenge to produce works on other aspects of the HGP and genomics, whether differing national and international projects or other genomics programs not discussed here. Our hope is that these multifaceted contributions will spur new waves of scholarships emphasizing the complexity, contingency, and controversy of the HGP and genomics that are simultaneously juxtaposed alongside its coherence, heterogeneity, and importance.

6. Postscript

As discussed above, this collection was initially developed before the COVID pandemic, reviewed and revised under gradually lessening pandemic conditions. More recently, it was decisively impacted by the “restructuring” of the NIH in 2025 by the Trump administration. In 2021 and continuing into 2022, in response to the Biden administration’s emphasis on anti-racism and equity in science, and the decisive role of the Black Lives Matter movement, the NHGRI and the History of Genomics Program took steps to address the persistence of eugenics, scientific racism, and ableism in genetics and genomics, and to confront the histories of these ideologies and practices through a series of meetings and public engagements, which are available on the NHGRI YouTube page. These events underscored the continuity of many eugenic ideas and practices through the era of the HGP, especially in the context of prenatal screening and testing (Parens and Asch 2003) and efforts to tie genetic and genomic data to complex social and behavioral traits. As Camisha Russell and other scholars have noted (Russell 2018), assisted reproductive technologies have functioned in the genomic era to reproduce biological, reductive notions of race and thereby also reproduce the episteme of historical and present-day eugenics movements. Although these elements are discussed in this collection, Donohue’s continuing research into geneticists’ (including Watson’s) post-war defenses of eugenics and their connections with genomics, and his more recent scholarly work on disability and genetics, would have been stronger components if this volume had been conceptualized in 2022. A full discussion of these topics is under preparation by Donohue and Michael Rembis, of the University at Buffalo. As importantly, this volume has been unable to fully take into account the continuing dilemmas of social and behavioral genomics, its continuity with behavior genetics, and the complex legacies of eugenics and scientific racism in these inquiries. These new efforts in social and behavioral genomics now attempt to link SNP variation to occupational and educational attainment (Lee et al. 2018), among other complex social dynamics. Although the gains from these studies are extremely small and their utility highly contested, they have already been used to fuel genocidal violence as well as antisemitism in the United States (Wedow et al. 2022). This raises another possibility that was not considered when this introduction was first written—even though the genomics community has taken steps to address legacies of racism and eugenics, this alone cannot and does not prevent the use of genetic studies by malicious actors. Additionally, the summer of 2024 saw an important, wide-ranging discussion around the ethics and informed consent process of the largest BAC library used in sequencing the human genome by Undark and many other media outlets, as discussed by Lotempio and colleagues (LoTempio et al. 2025).

Perhaps most consequently, as this volume was in production, the Trump administration terminated the History of Genomics Program in April 2025 and forced into retirement key NHGRI staff, including Eric Green. Christopher Donohue is now a researcher at the Institute for Clinical and Translational Science (ICTS) at University of California, Irvine, and a member of the LoTempio Lab. As the archive remains open for access, it is hoped that this volume spurs a larger community of researchers from diverse disciplines to apply for access and undertake work in the digital archives.

Notes

We are grateful to each of the volume’s authors for their contributions, which distinctively shaped our own thinking as we wrote the introduction, and patience as we assembled the pieces to our complicated puzzle over an extended period of time under unprecedented circumstances. We also appreciate the critical remarks and helpful suggestions from two anonymous referees that improved the chapter. Several of the chapters in the volume were substantially completed by 2018 and, due to a variety of factors, could not be systematically updated. The contributing authors hope that this will encourage other scholars to delve deeper into the history of the Human Genome Project and other genomics initiatives, as well as the NIH more generally.

1. 1. A slightly more technical definition of the “draft sequence” of the genome is necessary here. The draft sequence covers around 96 percent of the euchromatic—the more transcriptionally active and less densely packed—part of the genome, comprising around 94 percent of a human genome.

2. 2. There remain contentious issues with respect to Celera’s use of the public Bacterial Artificial Chromosome (BAC) libraries and its dependence on the public project’s data for sequence assembly (Cozzarelli 2003). BACs are engineered DNA molecules that facilitate cloning large amounts of DNA, often for the purpose of sequencing.

3. 3. Several provocative analyses emphasize deep continuities in this regard (e.g., McMahon 2019, 2020). However, these analyses are hampered by difficulties related to inaccurate characterizations of pertinent details of the science and problematic assumptions about the ubiquity of ideologies in diverse communities of inquiry.

4. 4. See T2T Consortium, https://sites.google.com/ucsc.edu/t2tworkinggroup.

Interview and Archival Materials

1. Interview with Deanna Church. Original Interview date May 3, 2019. Available online at https://www.genome.gov/player/ukjV3sbQxTg/PL1ay9ko4A8sk0o9O-YhseFHzbU2I2HQQp
2. Interview with Elke Jordan. Original Interview date January 28, 2014. Available online at https://www.genome.gov/player/pHCq2L5-2hs/PL1ay9ko4A8sk0o9O-YhseFHzbU2I2HQQp
3. Interview with Eric Lander. Original Interview date March 20, 2019. Available online at https://www.genome.gov/player/wtasZh9qy5M/PL1ay9ko4A8sk0o9O-YhseFHzbU2I2HQQp
4. Interview with Charles Rotimi. Original Interview date November 19, 2015. Available at https://www.youtube.com/watch?v=H6tg-48aVP4
5. NHGRI Internal History Archive. Scanned Francis Collins Box 1373–2, folder 029.
6. NHGRI Internal History Archive. Scanned Francis Collins Files, Box 7077, folder 27.
7. NHGRI Internal History Archive. Scanned Francis Collins Files, Box 7077, folder 14.
8. NHGRI Internal History Archive. Scanned Files btw NHGRI & Outside Orgs 1984–1988/15 Ad Hoc Cmte On CF Carrier Screening 1991–1992, folder 002.
9. NHGRI Internal History Archive. Scanned Box 1307, folder 017.
10. NHGRI Internal History Archive. Scanned Box 035, folder 017.
11. NHGRI Internal History Archive. Scanned Mark Guyer Collection, Box 13, HapMap folder.
12. NHGRI Internal History Archive. Scanned Francis Collins Files, Box 7108, folder 004.

References