https://orcid.org/0000-0001-9395-7001
Citation: Magaña P, Gopalasingam P, Fleming JR, Kovalevskiy O, Žídek A, Ebenezer TE, et al. (2026) Towards globally equitable bioinformatics adoption. PLoS Biol 24(7): e3003839. https://doi.org/10.1371/journal.pbio.3003839
Published: July 1, 2026
Copyright: © 2026 Magaña et al. 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: Google DeepMind (contract with EMBL to the teams of SV and CB) supported this work by funding the AlphaFold Summit and by contributing to its design and delivery.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AI, artificial intelligence; HPC, high-performance computing; IDRC, Infectious Diseases Research Collaboration; LMICs, low- and middle-income countries.
Advances in bioinformatics and artificial intelligence (AI) have reshaped the life sciences, enabling large-scale data analysis, predictive modeling, and integrative approaches across disciplines [1,2]. Many contemporary bioinformatics tools and AI models are released as open-source resources, with publicly accessible code repositories, pre-trained models, and web-based interfaces. This widespread availability has reinforced the expectation that technical barriers to participation have largely been removed; however, experience across multiple global locations reveals a persistent disconnect between open access and open utility. Analyses conducted by Google DeepMind on AlphaFold citation data using OpenAlex [3] and national diagnostics show that researchers in low- and middle-income countries (LMICs) remain underrepresented in the development and downstream application of advanced bioinformatics tools, even when these tools are freely available. These patterns are consistent with long-standing disparities in computational infrastructure, training capacity, and institutional investment [4,5]. Ensuring equitable adoption of bioinformatics tools will therefore require more than making the tools themselves open access (Box 1).
Box 1. Priorities for equitable adoption of advanced bioinformatics tools.
1. Enable shared infrastructure and low-barrier computational access.
For example: foster collaborations with cloud computing providers, form partnerships with high-performance computing centers or well-resourced institutions, and develop training materials tailored for offline use and low-bandwidth environments.
2. Invest in scalable Train-the-Trainer and mentorship models.
For example: create training resources suitable for various delivery methods, such as in-person, online, and hybrid formats, and make them openly available with guidance on how to use them.
3. Prioritize language accessibility and contextual relevance.
For example: translate training materials into the local language and provide use cases relevant to areas of direct interest to the trainees.
The advent of AI has profoundly transformed bioinformatics applications, transitioning many from rigid, deterministic local pipelines to scalable, data-driven computations that demand far greater computational resources. AlphaFold 2 exemplifies this broader structural challenge. Its release, together with the AlphaFold Protein Structure Database, made large-scale predicted structures broadly accessible in principle [1,6], yet routine use and methodological innovation remain concentrated in settings with established computational infrastructure and specialist expertise [4,7,8]. Similar patterns have long been documented across genomics, transcriptomics, metagenomics, and public health bioinformatics, indicating that these challenges are systemic and long-standing rather than specific to any single tool or technological generation [9–11].
The arguments presented here draw on both published analyses and structured discussions held during the AlphaFold Education Summit, which brought together educators and researchers from multiple regions, with strong representation from LMICs (see the full report here). Across both literature and summit discussions, access to computational infrastructure was described as a central limitation. Many contemporary bioinformatics methods depend on high-performance computing (HPC), graphic processing units (GPU), large-scale storage, and reliable internet connectivity, yet in many LMIC contexts, computational resources are decentralized and siloed within selected institutions, limiting broader accessibility [5,11]. These limitations are further compounded by unreliable power supplies and low-bandwidth internet connectivity. Under these conditions, cloud-based solutions, despite occasionally being designed to minimize reliance on computing infrastructure (e.g., the AlphaFold Database) and promoted as accessible tools, may remain difficult to use reliably and sustainably, particularly given their ongoing cost implications. In these contexts, the mere existence of sophisticated software, and even access to it, provides limited practical benefit.
Training limitations further compound these constraints. Regional capacity assessments and mentorship program evaluations consistently report shortages of trained bioinformaticians, instructors, and infrastructure administrators in resource-limited settings [4,5]. Training initiatives are frequently short-term and externally driven, providing limited opportunities for progression or local ownership. Evidence from mentorship-based programs and long-standing regional networks indicates that one-off workshops rarely yield sustained expertise, whereas longer term, mentored approaches are effective in building local capacity to sustain these efforts, thereby fostering independent analytical capability and local leadership [4,10].
Language and contextual accessibility further influence participation, and are often under-recognized barriers. Evaluations of training initiatives in LMICs emphasize the importance of locally relevant datasets, multilingual materials, and adaptive adult-aimed design [12]. Summit participants echoed these findings, highlighting that sustained engagement depends not only on technical instruction, but on researchers’ ability to apply tools independently within locally relevant research contexts. Furthermore, securing adequate and sustained funding for training programs, essential infrastructure, and long-term community-building initiatives remains a significant concern, as it threatens continuity and contributes to cycles of capacity loss.
Drawing on both summit deliberations and established regional initiatives, five interrelated priorities emerge for equitable adoption of advanced bioinformatics tools (Box 1). Central among these is resource-aware design: developing training materials suitable for low-bandwidth environments, including pre-configured notebooks and locally deployable workflows, and fostering collaborations with HPC centers and cloud providers to mitigate computational constraints. Ensuring that host institutions possess baseline hardware capacity is equally important.
Another critical element is scalable training and mentorship networks. A core feature of the AlphaFold Education Summit was a Train-the-Trainer model [13], designed to strengthen local training capacity by integrating technical instruction with andragogical development and community-building strategies. This approach aligns with earlier regional initiatives such as BioStruct-Africa [8], CABANAnet in Latin America [14], and APBioNet in the Asia–Pacific region [9], which demonstrate that capacity building is most effective when organized around communities of practice rather than individual platforms. Such networks enable peer mentorship, shared infrastructure, and engagement with science policy across evolving technological landscapes [10,14]. Such models, when embedded within longer term mentorship frameworks, support skill retention, independence, and the development of local research leadership [4,7,10]. These approaches move beyond episodic instruction toward sustained institutional capacity.
Language accessibility and contextual relevance are prerequisites for inclusion. The careful development of multilingual training materials, validated by local experts, and the grounding of instruction in locally relevant research questions, strengthen learner confidence and facilitate integration of new tools into existing research programs. Finally, institutional and policy integration are essential for sustainability. Embedding bioinformatics training within university curricula, aligning initiatives with national research strategies, and supporting stable career pathways are paramount to retaining expertise and avoiding cycles of capacity loss [8]. Collaborative funding models, including institutional consortia, industry partnerships, and philanthropic engagement, may further enhance resilience.
The rapid expansion of AI-enabled and resource-intensive bioinformatics tools makes it clear that democratization cannot be assumed to follow from open access alone. AlphaFold illustrates how even widely celebrated open access tools can reproduce existing inequities when infrastructure, training, and institutional support remain unevenly distributed. Ensuring equitable participation, therefore, requires deliberate, sustained investment in shared infrastructure, long-term training networks, language accessibility, institutional integration, and global communities of practice. Without such commitments, the benefits of technological innovation risk remaining concentrated in already well-resourced settings. Treating equity as integral to scientific infrastructure, embedded within funding decisions, institutional planning, and capacity building policy, offers a path toward a more inclusive and globally representative research ecosystem.
Acknowledgments
We would like to thank Barbara Etzi at EMBL-EBI for her superb organization, which brought the AlphaFold Education Summit to successful fruition. We would also like to thank all of the funding bodies that enabled the participation of summit participants through support and travel awards. Google DeepMind (contract ID IC26957 to RG); Wellcome (grant ID 228142-Z-23-Z enabled participation of TEE; grant ID 222999/Z/21/Z to EN); CONAHCYT-SECIHTI CF-2023-I-1653 (to YIB-M); Council of Scientific & Industrial Research (CSIR) grants FBR070301 and OLP0028 (to SC); African Society of Human Genetics (to FGF); a DBT/Wellcome India Alliance intermediate fellowship (IA/I/21/2/505928, to MG); NIH (contract ID U2RTW012116) to SK; a DBT for MK Bhan Fellowship (to SKM); Agencia Nacional de Investigación y Desarrollo (ANID, Chile) through the FONDECYT Regular program (grant 1211731), the FONDAP program (grant 1523A0008), and the Anillo program (grant ATE220016); by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, Brazil) through INCT-DigiSaúde (grant 408775/2024-6) and a Research Productivity Fellowship (grant 314606/2026-2); and by the Plataforma ICAI of the Universidad de Chile (grant FIU-138310) (to V M-C); UCR C5027 and C4604 (to JAM-M); GlaxoSmithKline, Novartis, and the South African Medical Research Council (to HO); and Sydney Brenner Charitable Trust (to DT).
References
- 1. Kovalevskiy O, Mateos-Garcia J, Tunyasuvunakool K. AlphaFold two years on: Validation and impact. Proc Natl Acad Sci. 2024;121(34):e2315002121.
- 2. Jiang J, Li Y, Cao S, Shan Y, Liu Y, Fei T, et al. Artificial intelligence in bioinformatics: a survey. Brief Bioinform. 2025;26(6):bbaf576. pmid:41206113
- 3. Priem J, Piwowar H, Orr R. OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts. arXiv. Available from: http://arxiv.org/abs/2205.01833. 2022. Accessed 2025 May 28.
- 4. Jjingo D, Mboowa G, Sserwadda I, Kakaire R, Kiberu D, Amujal M, et al. Bioinformatics mentorship in a resource limited setting. Brief Bioinform. 2022;23(1):bbab399. pmid:34591953
- 5. Tibiri EB, Boua PR, Soulama I, Dubreuil-Tranchant C, Tando N, Tollenaere C, et al. Challenges and opportunities of developing bioinformatics platforms in Africa: the case of BurkinaBioinfo at Joseph Ki-Zerbo University, Burkina Faso. Brief Bioinform. 2025 Jan 1;26(1):bbaf040. https://doi.org/10.1093/bib/bbaf040 pmid:39899597
- 6. Fleming J, Magana P, Nair S, Tsenkov M, Bertoni D, Pidruchna I, et al. AlphaFold protein structure database and 3D-beacons: new data and capabilities. J Mol Biol. 2025;437(15):168967. pmid:40133787
- 7. Nji E, Cramer KC, Rüffin NV, Fofana FG, Heiba W, Sankhe S. Leveraging AlphaFold for innovation and sustainable health research in Africa. Nat Commun. 2025;16(1):1334. https://doi.org/10.1038/s41467-025-56545-y. pmid:39905054
- 8. Nji E, Moumbock AFA, Cramer KC, Rüffin NV, Davis J, Asojo OA, et al. Supporting structural biologists in Africa requires resources and capacity building. Nat Struct Mol Biol. 2024;31(12):1814–5. pmid:39613973
- 9. Khan AM, Tan TW, Schönbach C, Ranganathan S. APBioNet-transforming bioinformatics in the Asia-Pacific region. PLoS Comput Biol. 2013;9(10):e1003317. pmid:24204244
- 10. Aron S, Chauke PA, Ras V, Panji S, Johnston K, Mulder N. The development of a sustainable bioinformatics training environment within the H3Africa bioinformatics network (H3ABioNet). Front Educ. 2021;6:725702. 10.3389/feduc.2021.725702
- 11. Aruhomukama D, Galiwango R, Meehan CJ, Asiimwe B. Enhancing genomics and bioinformatics access in Africa: an imperative leap. Lancet Microbe. 2024;5(5):e410–1. pmid:38281497
- 12. Moore B, Carvajal-López P, Chauke PA, Cristancho M, Dominguez Del Angel V, Fernandez-Valverde SL, et al. Ten simple rules for organizing a bioinformatics training course in low- and middle-income countries. PLoS Comput Biol. 2021;17(8):e1009218. pmid:34411091
- 13. McGrath A, Champ K, Shang CA, van Dam E, Brooksbank C, Morgan SL. From trainees to trainers to instructors: sustainably building a national capacity in bioinformatics training. PLoS Comput Biol. 2019;15(6):e1006923. pmid:31246949
- 14. Campos-Sánchez R, Willis I, Gopalasingam P, López-Juárez D, Cristancho M, Brooksbank C. The CABANA model 2017–2022: research and training synergy to facilitate bioinformatics applications in Latin America. Front Educ. 2024;9. Available from: https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2024.1358620/full