Abstract
Recent accounts of “the biological” emphasize its thoroughgoing transformation. Accounts of biomedicalization, biotechnology, biopower, biocapital, and bioeconomy tend to agree that twentieth- and twenty-first-century life sciences transform the object of biology, the biological. Amidst so much transformation, we explore attempts to stabilize the biological through standards. We ask: how do standards handle the biological in transformation? Based on ethnographic research, the article discusses three contemporary postgenomic standards that classify, construct, or identify biological forms: the Barcoding of Life Initiative, the BioBricks Assembly Standard, and the Proteomics Standards Initiative. We rely on recent critical analyses of standardization to suggest that any attempt to attribute a fixed property to the biological actually multiplies dependencies between values, materials, and human and nonhuman agents. We highlight ways in which these biological standards cross-validate life forms with forms of life such as publics, infrastructures, and forms of disciplinary compromise. Attempts to standardize the biological, we suggest, offer a good way to see how a life form is always also a form of life.
Introduction
The object of biology, the biological, is changing. There is some agreement on how this has happened. Transformations in the life sciences have been long discussed, especially in the context of molecular biology and its emphasis on DNA as code (Kay 2000), but also in accounts of biomedicalization (Clarke et al. 2003; Keating and Cambrosio 2003), biotechnology, the “bioeconomy” (Organization for Economic Cooperation and Development [OECD] 2009), “biocapital” (Rajan 2006; Cooper 2008; Waldby and Mitchell 2006; Fortun 2008), or biopower (Rabinow and Rose 2006). There is less agreement on how to study change in the biological. In her work on the history of cells in twentieth-century biology, Hannah Landecker analyzes change in what counts as “biological” through the doing of biology. She advocates “keeping an eye on practical, protocols, methods, technique, touch, or infrastructure” in order to catch sight of “the ways in which work on some life (nematodes, insects, yeast) reshapes human life by introducing systematic change into biological existence” (Landecker 2007, 234). More generally, based on several studies of “limit biologies”, Stefan Helmreich argues that “life forms and forms of life not only inform each other … but may be impossible to disentangle” (Helmreich 2011, 693), the implication being that perforce we should study entanglements. As Helmreich explains, “by ‘life forms,’ I mean those embodied bits of vitality called organisms, variously apprehended as ranged into species (durable, but changeable genealogical kinds) or as sorted into types occupying spaces of physical, metabolic, or ecological possibility (e.g., photosynthesizers, deep-sea dwellers) … When I write of ‘forms of life,’ I mean those cultural, social, symbolic, and pragmatic ways of thinking and acting that organize human communities” (Helmreich 2009, 6). In both Landecker’s and Helmreich’s view, life forms and adjacent forms of life reshape each other, so we would do well to learn to track and map their reshapings.
The transformations associated with biomedicalization, biotechnology, the bioeconomy, biocapital, or biopower rely implicitly or explicitly on standards and standard making to fix or stabilize something about life in the quest for good health, innovation, the market, profit, objectivity, regulation, biodiversity, or the planet. But the essential thing that a standard needs—an unchanging attribute—is, it seems, precisely what the biological essentially lacks. Every putative attribute—species, gene, population, individual, disease—seems to deform sooner or later. The paradox of standards for living things, whether they are patients, mice, cells, citizens, or ecosystems, is that fixed or constant forms are elusive.
Sociological work on standards has articulated versions of this paradox. In a sophisticated analysis of how biomedical standards govern lives, Laurent Thévenot asks what happens when standards encounter a plurality of lives. In a dense but insightful discussion, he writes: “standardization provides a guarantee by attributing properties to standardized objects, while all the while the very notion of engagement accounts for guarantees that rest on a dependency between agent and environment that goes against any such attribution” (Thévenot 2009, 807). An engagement for Thévenot comprises “an arrangement of the material environment” and as well as some value components or “a quest for a good” (p. 802). Regimes of engagement set up material environments to offer “a pledge (gage)” for a good of some kind (conventional public good, a feeling of ease, satisfication). Engagements have different epistemic, technical, emotional, aesthetic, political, social, and economic architectures. But in all engagements, the dependencies between agent and environment tend to resist attribution of fixed properties (of which “physico-chemical qualities are held as an ideal” (p. 805)). Put differently, processes of standardization seek to bootstrap on the very thing—engagements or “agent-environment dependencies”—that elude fixity. The implication of the sociological work on standards is that to the extent that recent biotransformations—biomedicine, biotechnology, the bioeconomy, biocapital, biopower—depend on standards in their pursuit of goods such as care, biodiversity, profit, civic order, and regimes of engagement will intrinsically tend to vary, resist, or disrupt the stable attribution required for standardization. This kind of circularity inherent in standards, if it matters, must in turn affect any putative transformation of “the biological.”
Large-scale, data-intensive, information system-reliant biologies such as genomics, proteomics, and synthetic biology exacerbate the difficulties of locating fixed attributes in living multiples. These heavily equipped and computationally sophisticated knowledge enterprises invest heavily in standards. In this article, we discuss three cases of what we call “cross-validating standardization,” all located in post-Human Genome Project life sciences. 1 The attempts to standardize species classification in the Barcoding of Life Initiative (BOLI), to standardize the construction of DNA sequences, so they can be assembled as biological parts in BioBricks, and to standardize the identification of proteins in the Proteomics Standards Initiative (PSI) are associated with different material arrangements and seek different ends or goods.
In attempting to standardize biological things in these high-throughput ways, what happens to the biological? The taxonomic, synthetic biology, and proteomics standards we discuss all accentuate uncertain, wavering shifts in the relation between forms of life and life forms. It is in “equipped communities” (Thévenot 2009, 796) that value high-throughput, high-ambition knowledge production, standardization itself undergoes deformation. As we will see, as they classify, construct, and identify, BOLI, BioBricks, and PSI each find themselves caught in tensions between the diverse life forms the standard seeks to bring to order and the multiple forms of life or regimes of engagement in which the standards are made. This is nothing new to science, technology, and society (STS) scholars, who have long emphasized entanglements and attachments as the everyday order of things. But the cases highlight something different as well. As they classify, construct, and identify life through a well-defined locus or site—a gene, a DNA restriction site, and a set of protein-related measurements—they simultaneously amplify the regimes of engagement. They explicitly or inadvertently invoke, for instance, publics who can add something that the scientists or engineers cannot. Public participation, democratization, or simply open cooperation will, it is hoped, guarantee the standards’ attribution of properties to living things. Validating publics are examples of the agency-environment dependencies that emerge when narrowly specialized technoscientific cultures become broadly ambitious. Similarly, we will see that as they “pass to the lives of elementary things” such as fungi, cells, or proteins (Thevenot, 2009, 805), standards generate forms of infrastructural liveliness. They invest heavily in formal material arrangements such as databases, registries, catalogs, and web interfaces that superimpose other cross-validating regimes of engagements with divergent consequence.
Classifying: Differences in the BOLI
The ordinary problems of taxonomy (the science that is concerned with ordering plants, animals, and microorganisms into groups based on their shared characteristics) encompass a specific kind of metaphysics—a metaphysics that also holds together the transmission of knowledge and a community of knowers through the creation of singular designators for things that are innately variable, and that shift over time (Daston 2004). Standards are vital in this metaphysics that binds together life and politics, ontology, and accountability (Neyland 2008). A recent effort aimed at bringing genomics to bear on the taxonomic sciences—the BOLI—involves the making of new standards for taxonomy and hence global biodiversity knowledge and ultimately protection.
Led by Paul Hebert from the University of Guelph, Canada, BOLI emerged, in 2003, from the increasingly refined use of “micro-genomic” techniques for classifying and identifying species. BOLI, in effect, realized a taxonomic dream. If only, went the dream, each living species on earth embodied one single and unique “character,” the variations within such a character could be exposed to judge biologically reproducing kinds apart. Such a fixed attribute would simplify and create unity from the messy proliferation of taxonomic approaches currently drawn upon to classify and name living things (Ridley 1986; Hull 1988). The accompanying dream was that this simplification would simultaneously open up taxonomic practice and knowledge to ordinary citizens and in so doing pave the way for a shared responsibility for the protection of global biodiversity.
Hebert and colleagues were looking for a genetic “universal marker” found in all living organisms. It would index diversity yet be amenable to the sequencing and visualization techniques necessary for the marker’s wider use. The “magic bullet” found by Hebert and colleagues was 648 base pairs of the mitochondrial cytochrome oxidase I (COI) gene, a gene fragment that exhibits differentiation along existing understanding of animal species boundaries, and that can be translated, using sequencing, informatics, and digital technologies combined, to form a visual representation that looks just like a consumer product “barcode.” Hebert and team announced in 2003: “‘genetic barcodes’ are embedded in
The naturalizing and universalizing claims emanating from Guelph and rapidly disseminating globally from 2003 onward, belied the craftwork invested in the fabrication of the genetic marker, COI. As in other areas of genomic science, such crafting involved a careful calibration between organisms, reagents, algorithms, sequencing, and data management technologies. These components of a DNA-based barcode were, under BOLI, to be ratcheted up to a large-scale, high-throughput, species searching and identifying initiative (BOLI itself) that would enroll taxonomists globally in the quest to barcode all species on the planet.
Barcoding Lives
DNA barcoding’s promise to co-order life forms and form of global public life was disappointingly short-lived. It quickly became apparent that the extraordinary attributes of COI for animals were not shared by plants, fungi, and protists. COI simply did not reveal enough base-pair variability to discriminate between all plant species in the way it appeared to do for animals. If life was internally standardized, standards it seemed were plural.
During 2005-2008, at least nine institutions were involved in selecting “plant barcode” from twelve candidate genetic regions. The apparent seamless naturalization of the animal barcode contrasted greatly with the many contingencies of fabricating, testing, and finally selecting a plant barcode. Such contingencies potentially disrupted the very core of life’s standardization predicated on a shared attribute of all living things. From the long and complex search to find the plant barcode a compromise emerged. It was a compromise between the call, from some, for pragmatism, and the demand, from others, to fully recognize plant life’s infinite variety, specificity, and genomic dynamism. Advocating an easily retrievable and robust gene and referring to the fact that most public “users” of DNA barcoding would not need to be privy to the esoteric side of taxonomic science, one participant stated “The FBI are not interested in historic kinship!” Conversely, and arguing that the benefits of ease and replicability should not outweigh the very purpose of taxonomy, another participant exclaimed, “we should not sacrifice accuracy for universality!” Eventually, a “composite” barcode was agreed upon combining the gene region rbcL as a universalizing and safe “anchor” and a further region, matK, as a high-resolution marker able to discriminate at the species level. All standards are mixed, as STS scholars have shown (Bowker and Star 1999, 39), but we can highlight two further issues in these negotiations. First, some life forms—plants—seem to resist attempts to identify their differences at a genetic level, thus casting doubt on the very possibility of bringing these kinds of standards to life (Hollingsworth et al. 2009). Second, the ascendant role of bioinformatics in the decision-making process foregrounded some of the “ontological” problems which have wracked barcoding from the outset (i.e., what is a species?). One bioinformatician stated that algorithmically mediated approaches to species discrimination on the basis of sequence divergence amounted to “no more than a probability statement.” Here the attribution of a property to a thing—the key move in making a standard—has had to traverse increasingly long chains of inferences and imputations associated with DNA sequencing alignment.
A Data-gathering Public = “Enough Rope to Hang Ourselves by”
The barcoders also sought to “democratize” taxonomy through a new global data-gathering public. They were crafting (imagined) public user communities who would require the new services that taxonomy could now provide. The “democratization of taxonomy” (Costa and Carvalho 2007; Holloway 2006), via barcoding, came to take on an important and virtuous status within BOLI in the mid 2000s. Standardized technoscience was coimagined with standardized publics (Ellis, Waterton, and Wynne 2009). These imagined publics were partly newly envisaged “users” of a genomically inspired taxonomy—those who needed quick and easy identification of natural species, such as pest control officials and border agency units, aviation authorities, regulators of species trafficking and fisheries, and so on. But this imagined public also included an inspirational vision of the “unconverted majority”—a nebulous public deemed to be bioilliterate, sadly not biophilic (i.e., nature-loving), yet considered potential beneficiaries of the efforts by taxonomists to identify and archive in accessible digital form all species on the planet as fast as possible, and ideally before the majority of them become extinct (Janzen 2004b). These supplementary publics, in both their mundane (border control) and their more idealized (global bio-illiterate yet potentially biophilic) forms, began to permeate the very construction of the bio-informatic object in play, in this case the DNA barcode (see also Cambrosio et al. 2009). They also come together in the Barcode of Life Database (BOLD). BOLD was designed as a globally accessible “workbench” for the processing and archiving of barcoding data and constructed in such a way as to enforce “clean,” replicable barcoding practices. Specified numbers of samples, for example, were needed to underpin each single barcoding entry in the BOLD database. Every entry of barcoding data would be ranked according to whether these and other criteria had been met, with the highest ranking procedures, the “gold standard,” being the only level of quality to allow an entry to be endowed with the denominator “BARCODE DATA. BOLD, in effect, was a material arrangement designed to elicit public witness of standards.”
In BOLD, as one commentator suggested, barcoders had created “enough rope to hang themselves by.” The problem, as this commentator saw it, was that the standards for data processing and data entry required of barcoding practitioners had been set too high. Some five years after the establishment of BOLD, concerns were raised that the number of “BARCODE flag” records in the database was rather few (C. 400,000 in 2008), that only 1/8 of these were publicly accessible, and that many of them had not, in actual fact, met the barcode standard despite their having been given the BARCODE flag. As one BOLD expert commented: “There’s a big disconnect between the data standard and the community. We need to elevate the minimum amount that people think they need. BOLD is forced now to lower the bar, so that people can use the tool. We need to tread a line between the usability of BOLD and reaching the DATA standard.” 2 The idea of lowering of the standard to facilitate the achievement of BOLI’s original aim highlights the radical contingency and mutability of BOLD data standards. It shows the delicate trade-offs that were implicitly and sometimes explicitly being made between “high” BOLD standards, on one hand, and a full and lively global database consisting of new species barcode records, on the other.
Standards for classifying life faced life forms that wriggled around the categories, attributes, or properties that could only be fixed by long chains of statistical attribution and forms of life—publics—who were both needed yet never gave quite enough to weave the delicate metaphysical aggregate of life forms and forms life amid legion species extinction.
Constructing: BioBricks
Around the same time as BOLI, BioBricks originated in a computer science and artificial intelligence laboratory led by the computer scientist Tom Knight at the Massachusetts Institute of Technology (MIT; Knight 2010a). Named with explicit reference to Lego® bricks, BioBrick™ standard biological parts are DNA sequences encoding particular functions, designed for easy assembly into genetic circuits. They encapsulate an MIT-initiated ambition to apply engineering principles to the design and construction of new biological systems. Within a few years, BioBricks symbolized the pragmatic yet speculative ethos of the growing discipline of synthetic biology (e.g., Brent 2004; Endy 2005; Heinemann and Panke 2006). BioBricks standardize the interfaces of DNA sequences to allow them to be joined together easily. Like the COI mitochondrial DNA at the heart of BOLI, BioBricks trade on what many biologists describe as the intrinsic or natural modularity of biological signaling and decision-making systems (Hartwell et al. 1999; Agapakis and Silver 2009; Lim 2010). Building on analogies with technical systems such as computers and electrical circuits (Andrianantoandro et al. 2006), BioBricks are meant lay the ground for a biological engineering “abstraction hierarchy.” The abstraction hierarchy assembles modular and interchangeable component “parts” into larger biological “devices” or “systems,” hosted in different organisms (or “chassis”) to carry out defined and predictable functions. BioBricks are intended to allow synthetic biologists to “build stuff” out of biological material with minimal concession to the variabilities, differences, and contingencies of cellular life.
Researchers in molecular biology often devote significant resources to the preparation of cellular or biochemical constructs and experimental environments tailored to particular biological questions. BioBricks proponents such as Drew Endy, Randy Rettburg, and Tom Knight regard this approach as time-consuming and unpredictable: “a practicing experimental biologist or biological engineer can easily spend around 50% of their effort manipulating the DNA just to produce the genetic material needed for an experiment” (Endy 2005, 452). Standardized BioBrick parts are designed to improve the efficiency and automation of DNA assembly, and to circumvent the “tedium and surprise” (Knight 2010b, 2) currently associated with the preparation of DNA constructs.
It is hard to know whether DNA assembly through the use of BioBricks makes a big practical difference, but many synthetic biologists are actively engaged in making standard biological parts. The pragmatism of the standardization agenda of synthetic biology supports plans and expectations about the future of biological engineering. Proponents of BioBricks use them to connect the growth of biological engineering with the promise of useful and economically profitable applications in areas as diverse as biomanufacturing, medical diagnostics and therapeutics, and environmental biosensing and remediation. Standards development is firmly aligned with aspirations about the utility, scaling up, and industrialization of synthetic biology. Progress in engineering biology is actively held back by the lack of common, standardized tools and technologies: “tremendous costs accrue owing to the lack of standards” (Endy 2005, 450). Standards will permit something comparable to the Industrial Revolution of the nineteenth century: “ … the standardization of pitch, diameter, and form of screw threads [provided] the infrastructure which allowed the industrial revolution to take off” (Knight 2010a, 2). Framed this way, standardization became an imperative, necessary to prevent ongoing wastage of biological potential. Like BOLI, it invokes broad participation: “can the reengineering of biology be coupled to the spread of tools and knowledge sufficient to improve the health of people and the environment worldwide?” asks Drew Endy (Billings and Endy 2010). The “black-boxing” of DNA assembly through the use of BioBricks is often presented as “democratizing” biotechnology, opening the door for new researchers, practitioners, hobbyists, and to enter the world of biological engineering.
Standard Assembly as a Form of Life
As products of a standard, BioBricks cross-validate engagement with plans and with markets. As plans, they bind the construction of life forms to heavily formalized biological material—DNA sequences—in order to support plans and design that can carry engineers into the future. By developing a standard process, BioBricks claim to transform the highly contingent knowledges, care, and skills required to assemble genetic constructs—what in Thévenot’s terms might a regime of biologically familiar engagement—into a planned engagement (p. 804). But this engagement also urges people to make and share BioBricks, and to coordinate their actions in a networked time space that looks more like a market than a laboratory. The attribution of a fixed property to biological parts aims to also ease their production. In isolation, BioBrick parts have little worth. They only count when combined in novel or useful biological devices. As in BOLI’s BOLD, the researchers at MIT (Knight and Endy) soon set up an online publicly accessible database, a repository of standardized, interchangeable biological parts, “The Registry of Standard Biological Parts” (partsregistry.org). Given a wiki-style registry of parts, it was easy for synthetic biologists and commentators to draw parallels with the quasi-distributed production of source code in open source software (Henkel and Maurer 2007). Open source production was also rendered more competitive by the rapidly growing BioBrick-based International Genetically Engineered Machine (iGEM) competitions held at MIT annually each year since 2004 (uno.igem.org). Finally, starting in 2006, the BioBricks Foundation, a “501(c)(3) public-benefit organization” stages and promotes ethico-legal-social framing of the promise of BioBricks under the rubric “biotechnology in the public interest” (BBF 2010).
Despite the rapid build-out of BioBricks in online registries, in a flourishing global student competition, in scientific laboratories, in scientific and media, and in an apparently high-profile public interest body promoting BioBricks, widespread consensus regarding the underlying standard for BioBrick parts proven elusive. Interestingly –but perhaps not unexpectedly –as this community has grown, it questioned and challenged fundamental aspects of BioBrick standards themselves. For example, the first BioBrick standard (known informally as the BBa standard, Knight 2010d) assembled DNA using restriction enzyme digestion followed by the ligation of DNA fragments. Within the growing community of practice, some see the BBa standard as technically flawed or as constraining the design space for synthetic biology. Like the original BOLI proposal, it turns out that this original standard is not appropriate for use across all “chassis” organisms (e.g., gram-positive and gram-negative bacteria, yeast, and mammalian cells) or for all possible purposes (including, e.g., modular protein engineering). Despite the pragmatic, real-world justification of standardized assembly, BioBricks effectively ran into the limits of biological ideas about modularity in biology. BioBricks as a form of life, as a way of bringing engineering to bear on biology, were meant to take something from the life forms—their modularity—and turn it to intentional ends by crafting modules that have purpose and function determined by humans rather than other living things. In practice, it seemed to be more challenging than originally anticipated to construct modules to work under any biological conditions.
In response to perceived shortcomings of the BBa standard, synthetic biologists working in laboratories across the US and Europe modified BioBrick standards. They sought to loosen constraints present in the original standard or to optimize BioBricks for different purposes (e.g., Phillips and Silver 2010; Knight 2010c; Peisajovich et al. 2009; Grünberg et al. 2010; Anderson et al. 2010). Importantly, some of the standards are incompatible with one another —for example, a BioBrick designed according to the BBa standard (Knight 2007) cannot be readily assembled with a BioBrick in the Silver standard format (Phillips and Silver 2006). Individuals or laboratories pursuing different goals or research questions started to create collections of incompatible parts (see the Joint BioEnergy Institute Registry, Berkeley). Fractionation of BioBricks into different registries dilutes the flow of interchangeable parts, and the hence basic plan for faster and more flexible innovation.
Any “community” may actually derail its own commonality in basing itself on standards. In trying to secure widespread support for the transformative vision of synthetic biology, its early proponents left the potential size and membership of this imagined community completely open. Its commonality was explicitly open-ended: “If synthetic biologists continue crafting tools that simplify genetic engineering, it will become much easier for anyone, regardless of training, to construct novel biological systems” (Billings and Endy 2008, emphasis added). But at present, the most visible users of BioBricks are undergraduate iGEM teams and Do it yourself (DIY)-biology groups. Academic consensus regarding appropriate standards for BioBrick exchange and interoperability is frustrated by differences across the research questions they wish to pursue, their model organisms of choice, and their visions for what the future of synthetic biology might hold. While this growing community might voice hopes about the use of standard components and tools, devising standards that achieve both the technical ambitions and the social goals set out for synthetic biology is far from trivial. This is a story still very much in flux; the relationship between the technical and the social, between standards development and community formation (its growth, composition, and norms), between life-form and form of life, is dynamic and multifaceted. At present, one might suggest that the widespread adoption of a single BioBrick design standard could actually destabilize the growth of the synthetic biology community as a form of life. More broadly, we could say that BioBricks mistake an older investment in the DNA form for life itself. This is something that any biological standard can easily do: mistake a form of life for a life form. Moreover, BioBricks show what can happen when standards in the service of planned engagement slip over into different regimes such as market-style coordination.
Identifying: Proteins in the PSI
In July 2010, the web site of the
In 2002, the Human Proteome Organization (HUPO, founded 2001) launched the international Human Proteome Project, comprising three tissue initiatives—the Plasma Proteome Project, the Brain Proteome Project, and the Liver Proteome Project—plus the PSI (Orchard, Hermjakob, and Apweiler 2003). HUPO’s PSI was perceived to be, and to some extent was, driven by the “new kids on the block”: relatively youthful bioinformaticians such as Henning Hermjakob and Rolf Apweiler whose practices were again aligned with a West Coast, open source ethos.
Analyses of the results from HUPO and other large-scale proteomics experiments found that even when the sample was the same, the proteins depended on how data were processed and analyzed. The profile of proteins discovered in a sample could be altered by changing the search algorithms used to assign peaklists to peptides, or those used to assign peptides to proteins, or by using a different reference database. Consequently, the identity of the peptides and especially the proteins in a sample was “inherently ambiguous” (Martens and Hermjakob 2007). Moreover, this ambiguity was unavoidable. The question was, where to draw the line? What level of ambiguity was unacceptable?
The PSI argued that standards were needed for data, so that proteomics data sets produced by proprietary experimental platforms and analysis pipelines could be combined and compared and reused by others for further research (Pedrioli et al. 2004). By the end of 2004, the PSI had developed a stable release of its data standard for mass spectrometry (Orchard et al. 2004). This standard not only standardized peaklist data, it also provided standardized formats for expressing the raw data and for describing the experimental metadata, including the instrument type and settings used, and the ways in which raw data had been processed to produce the peaklists. In addition to data standards, the PSI was developing reporting standards. Known as “MIAPEs,” standing for Minimal Information about a Proteomics Experiment, these were checklists of which features of a proteomics experiment should be reported with the data (Taylor et al. 2007). Over this same period, the European Bioinformatics Institute was building the PRIDE repository to house proteomics mass spectrometry data sets (Martens et al. 2005).
The core members of the PSI mass spectrometry working group included representatives of hardware and software vendors, developers of data management and visualization tools, and academic bioinformaticians. It lacked laboratory practitioners or journal representatives. Laboratory practitioners would have helped the working group to tailor the standards more closely to experimental settings. Although invited to join, it proved hard to engage them in designing data standards, or in the public repository the data standard was meant to support. Given their apparent lack of interest, how would they be persuaded to standardize their data and deposit it in public repositories?
In June 2004 Molecular & Cellular Proteomics (MCP), the leading journal in the field, published a set of criteria for reporting protein identifications determined by mass spectrometry. The following May it organized a workshop in Paris to develop a data reporting standard for all journals. The invited participants included research scientists, representatives of software providers, mass spectrometer manufacturers, journal editors, and bioinformaticians, but not the PSI per se. The Paris guidelines and the PSI’s reporting standard (MIAPE-MS) included similar checklists of metadata. The PSI responded to the Paris guidelines by pointing out overlaps between their respective checklists, and calling for their future harmonization. Through this cooperation, it argued, PSI’s software-friendly data standards would operationalize the technical, checklist elements of the Paris guidelines. 4 Rather than seeking to standardize experimental arrangements or analysis of experimental data, the PSI’s primary aim was to increase the potential for data reuse through routine deposition of minimally yet adequately described, standardized data sets in a public repository.
The PSI’s proposal was not taken up. Some of the authors of the Paris guidelines regarded the PSI’s reporting standards as merely technical, having little to do with questions of biological quality and validity. Viewed as dominated by bioinformaticians, the PSI seemed to lack the authority to set standards for scientific practice and publication. Standards for proteomics publications should come from biological laboratories, with hands-on practical experience in proteomics mass spectrometry. The revised versions of the Paris guidelines published the following year did not refer to the PSI, its data standards or its MIAPEs, and did not make public data deposition a requirement for publication (Bradshaw et al. 2006).
Fall before PRIDE—with the Aid of Nature, Science, a Dinosaur, and a Wizard
In 2007, an editorial in Nature Biotechnology entitled “Democratizing proteomics data” recommended authors deposit their data in a public repository, preferably PRIDE, before submitting their manuscripts (Anon 2007). In 2008, a similar editorial in Nature Methods additionally acknowledged the importance of data standards and praised the achievements of the PSI (Anon 2008). Further support for PRIDE came from another nonspecialist proteomics journal, namely, Science. In this case, the motivation was not data sharing but the need to make data available for independent validation. In 2007, Science published an article by researchers claiming to have used mass spectrometry proteomics to sequence proteins from the fossilized remains of bone marrow from the dinosaur Tyrannosaurus rex (Asara et al. 2007). Not surprisingly, the article attracted much skeptical correspondence. In response, the editor required that the lead author make the supporting data publicly available for independent evaluation. After searching Google for a suitable repository, he chose PRIDE because of “PRIDE Converter”—its newly developed, data entry tool. The dinosaur proteomic data set raised PRIDE’s profile and demonstrated its utility not just for data sharing. Five years after the Paris workshop, the 2009 revision of the MCP guidelines stipulated public data deposition as a condition of publication, with PRIDE Converter as a recognized repository.
The mutually reinforcing effects of pressure from journals plus easy data entry with PRIDE affected the growth in protein spectra. Journal editors, laboratory practitioners, and peer reviewers do not need to know that the PSI’s data standards are alive and well behind the scenes. The PRIDE Converter software automatically converts all the common data formats used in proteomics mass spectrometry into a common data standard (PRIDE XML) based on PSI’s standards (Barsnes et al. 2009). Furthermore, the PRIDE Converter not only makes data submission quick and easy but also standardizes data sharing with a “wizard” whose “next” button is not active until the required metadata fields in the current screen are filled in. Passing through the PRIDE Converter makes MIAPE-compliance not only the invisible “Obligatory Passage Point” (Callon 1986) for entry into PRIDE but for entry in the proteomics science community. More generally, this biological standard shows a convention that turns a somewhat blind eye to the life form, and the forms of experimental life. Rather than fixing on a property of the biological, it fixes on the variety of forms of experimental life in proteomics (machines, techniques, search algorithms, etc.). Trying to preserve the diversity of techniques through the almost purely formal convention of a data standard, it succeeds in identifying many parts of life.
Conclusion
We have asked how critical work on standards can help us explore and situate transformations in the biological. How do standards help us rethink the biological? The key argument we take from STS work on standards is that any attempt to fix a property is likely to occlude the dynamic agent-environment dependencies on which it relies. BOLI, BioBricks, and PSI illustrate the potent desires and frustrations encountered in standardizing the living in the interests of classification, construction, or identification. In each case, the cost of investing in standardization can be seen in considerable contortions in the practices of biologists, engineers, and informaticians. They shift between fixing on something in life forms (a gene, a stretch of DNA that can be used to cut and paste sequences, physicochemical property of proteins) and fixing on forms of life (a supplementary public, a network of BioBrick producers, and a common, mandatory data format). Slippages between fixing on life forms and fixing on forms of life are not accidental to biological standards. These slippages embody the agent-environment dependencies on which standards stand.
What happens to biomedicine, bioeconomy, biocapital, or “the biological century” if we understand transformations in the biological in terms of standards? The contrasting standards might help us envision a slightly different account of how the biological is changing. In the standards, we see utopian and pragmatic appeals to norms of democratization, moves toward shared facilitation of utility through access to data, or techniques going hand in hand with attempts to bind life forms. Every attempt, however, to locate something fixed in life or forms of life occasions the development of interfaces, protocols, registries, and equipment that imply expanded practices of identification, classification, and construction. While the databases and their web site portals are sometimes regarded as incidental, technical, or prosaic adjuncts to work on the biological, they are also recombinatory hotspots where cross-validating life form/form of life entanglements play out. In their terms of access, they instantiate publicness; in their sometimes labyrinthine, nested architectures, they channel the intricacies of biological specificity; in their own plasticity and transformability, they are vectors of tangled values, beliefs and desires concerning speed, control, and economy. Standards are one place, in short, where we see that there is no life itself apart from forms of life.
The different standards display degrees of awareness of the impossibility of disentangling forms of life and life forms. In what biologists, bioinformaticians, engineers, and others hold onto and what they let go in biological standardization, we glimpse the difficulties that contemporary biology experiences in coming to grips with its own shifting performances of the real, in the competitive conditions which it has to negotiate for survival. Each of them—BOLI, BioBricks, and PSI—names a debilitating diversity that has reigned for too long in identifying, classifying, and constructing life forms. But in negotiating trade-offs between ideal standards and do-ability, in finding a way of pledging material arrangements to a sought-after good, there are risks and responsibilities involved in creating standards. Perhaps, standards that stand at some distance from life forms—for example, PSI—can actually accept this responsibility most openly. Standards that heavily invest life form specificity—BOLI and Biobricks—encrypt this risk in plans focused on tightly bound attributes of the living.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the UK Economic and Social Research Council (ESRC) under the auspices of the Centre for Economic and Social Aspects of Genomics (Cesagen). Claire Waterton, Rebecca Ellis and Brian Wynne acknowledge the ESRC's funding under the “Taxonomy at a Crossroads 2006-2009” responsive mode project, 2002-2012.