EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Architectural good practices for reproducible benchmarking in protein machine learning

Julián García-Vinuesa, Diego Fernández-Villegas, Michelle Soto-García, Xavier Cadet, Mehdi D Davari, Frederic Cadet, Juan A Asenjo, Roberto Uribe-Paredes, David Medina-Ortiz

Frontiers in Bioinformatics · 08.09.2026 · 6

DOI: 10.3389/fbinf.2026.1925740 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Bioinformatics, online erschienen am 08.09.2026. Verfasst von Julián García-Vinuesa und 8 weiteren Autor:innen. Beteiligte Einrichtung: Departamento de Ingeniería En Computación, Universidad de Magallanes, Punta Arenas, Chile. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 92 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Reliable benchmarking in protein machine learning requires biological tasks, datasets, representations, execution conditions, and evaluation regimes to be explicitly defined and traceable. Protein-specific dependencies, including label semantics, negative-class construction, evolutionary relationships, similarity control, inferred structures, protein language model configurations, and potential pretraining contamination, can alter benchmark interpretation. This Perspective provides an operational, protein-specific synthesis of established reproducibility practices through an artefact-linked architecture. We define five core requirements for verifiable benchmarking, including deterministic curation, persistent and versioned artefacts, explicit interface contracts, reproducible execution, and transparent evaluation. Prediction provenance, uncertainty, calibration, and explainability are treated as an additional decision-readiness layer. These requirements are translated into minimum records, provenance links, and verification checks. An antimicrobial peptide classification case study demonstrates their application to alternative task formulations, negative-class definitions, dataset compositions, and partitioning strategies within a controlled workflow.

Schlagwörter

Stichwörter der Autor:innen

FAIR workflowscomputational infrastructuregood practicesinteroperabilityprotein machine learningreproducible benchmarkingverifiable workflowsworkflow provenance

Autor:innen und Institutionen

  • Departamento de Ingeniería En Computación, Universidad de Magallanes, Punta Arenas, Chile.
  • Dartmouth College, Hanover, NH, United States.
  • Department of Bioorganic Chemistry, Leibniz Institute of Plant Biochemistry, Halle, Germany.
  • AI for Biologics, PEACCEL, Paris, France.
  • Departamento de Ingeniería Química, Biotecnología y Materiales, Universidad de Chile, Santiago, Chile.

Zitieren

APA

García-Vinuesa, J., Fernández-Villegas, D., Soto-García, M., Cadet, X., Davari, M. D., Cadet, F., Asenjo, J. A., Uribe-Paredes, R., & Medina-Ortiz, D. (2026). Architectural good practices for reproducible benchmarking in protein machine learning. Frontiers in Bioinformatics, 6. https://doi.org/10.3389/fbinf.2026.1925740

Vancouver

García-Vinuesa J, Fernández-Villegas D, Soto-García M, Cadet X, Davari MD, Cadet F, et al. Architectural good practices for reproducible benchmarking in protein machine learning. Frontiers in Bioinformatics. 2026;6. doi:10.3389/fbinf.2026.1925740

BibTeX
@article{GarcaVinuesa2026,
  author = {García-Vinuesa, Julián and Fernández-Villegas, Diego and Soto-García, Michelle and Cadet, Xavier and Davari, Mehdi D and Cadet, Frederic and Asenjo, Juan A and Uribe-Paredes, Roberto and Medina-Ortiz, David},
  title = {Architectural good practices for reproducible benchmarking in protein machine learning},
  journal = {Frontiers in Bioinformatics},
  year = {2026},
  volume = {6},
  doi = {10.3389/fbinf.2026.1925740}
}
Kostenlos starten →