Copyright: © 2025 Sloan, Stenglein. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The authors' research is supported by the National Institutes of Health (R35GM148134 and T32GM132057 to D.B.S.) and the National Science Foundation (IOS-2048214 to M.D.S.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Nearly every corner of the sciences has been transformed by technologies that produce massive data sets, placing us in the era of “big data” [1]. This progress has been accelerated by specialized companies and core facilities that can provide much greater throughput and efficiency than individual research labs. For the purposes of this Perspective, we focus on DNA sequencing facilities, but the general points pertain to centralized data collection in areas such as proteomics, metabolomics, cryo-electron microscopy and structural biology, high-throughput screening and drug discovery, and isotope/chemical analyses.

As data generation becomes increasingly centralized and commoditized, we anticipate that the problem will worsen, with researchers being further removed from the production process and more accustomed to receiving data from fee-for-service contractual exchanges. Although it is commonplace for researchers to purchase reagents from commercial providers and simply reference the source without information about how the reagents were produced, we contend that treating sequencing facilities in a similar black-box fashion is problematic. The high-throughput technologies that are driving progress in genomics and other big-data fields are evolving too fast. There are numerous cases where identification of sequencing artefacts has overturned initial scientific conclusions and raised questions about whether other studies suffer from the same methodological problems [2–6]. It is especially important to record methodological information when facilities process samples and construct sequencing libraries, activities that are more complex than data generation from existing libraries. Mundane details such as the DNA polymerase and number of PCR cycles used for library amplification can determine the amount of bias in sequence representation [7]. More generally, even in the absence of any technical errors or artefacts, proper (re)interpretation of sequencing data sets can depend on the specifics of how they were produced [8–10]. For example, whether reads represent the coding or template strand (or a mix of both) in RNA-seq data sets depends on the methods used during library prep [11].

One challenge arising from centralized data generation is that researchers often lack direct experience with the methods and, thus, may not know what questions to ask. As such, obtaining key methodological information may require back-and-forth exchanges among researchers, facility personnel, and other experts in the field. Early efforts were made to establish standardized reporting requirements for high-throughput sequencing experiments, including both wet-lab and computational methods [12,13]. However, these requirements have never reached widespread adoption in the same way that, for example, the Minimum Information for Publication of Quantitative Real-Time PCR Experiments (MIQE) guidelines defined norms for reporting experimental methods for quantitative PCR projects [14]. In addition, requirements for submission to repositories such as the NCBI Sequence Read Archive (SRA) have often placed greater emphasis on (very important) metadata about biological samples than on methods for generation of the sequence data itself.

Although the ultimate responsibility for recording and reporting methodological information lies with researchers, all parties can contribute to ensuring reproducibility in outsourced data generation (Box 1). Simple steps that researchers can take include confirming from the onset of a project that sequencing facilities will provide protocol information and making sure they do so upon project completion. In our experience, facilities are usually happy to supply this information (with rare exceptions raising a big red flag), but it generally requires a proactive request. Researchers should establish lab policies to record those methods and report them in publications and data depositions, just as they would for work conducted in their own labs. Peer reviewers, publishers, and funders also have roles to play (Box 1) by insisting on full methods reporting during manuscript review and as a condition of funding. Data repositories like the NCBI SRA could enforce these policies by requiring that data submissions be paired with more complete methodological information.