{"doi":"10.3389/fmolb.2024.1540823","title":"Editorial: Revolutionizing life sciences: the nobel leap in artificial intelligence-driven biomodeling","abstract":"Within the research world, 2024 will be remembered as the year of Nobel Prizes for Artificial Intelligence (AI). The one for Physics, awarded to John Hopfield and Geoffrey Hinton for foundational discoveries and inventions that enable machine learning with artificial neural networks, has sealed the connection between physics and information science, now officially mating on a strongly interdisciplinary frontier field after over fifty years of fruitful interaction [nat24]. More specifically, connecting AI to biomolecular modeling relates to the Nobel Prize in Chemistry awarded to David Baker for computational protein design and to Demis Hassabis and John Jumper for protein structure prediction.Numerous statistics illustrate the influence of artificial intelligence in the field of biomodeling. An inquiry conducted in scientific literature databases employing AI-related keywords pertinent to the computer modeling of biomolecules yields approximately 120,000 results (approximately 6,000 results if the search is confined to the abstract, as illustrated in Fig. 1). The exponential rise observed starting from 2018-19 was the prelude to the Nobel, and approximately coincides with the appearance of the two software suites, AlphaFold [Senior et al (2019)] and RosettaFold [Humphreys et al (2021)], which implement the methods for proteins folding and proteins de novo design developed by Hassabis/Jumper and Baker, respectively.Receiving a Nobel Prize just a few years after the awarded research is quite rare, but certainly not accidental. The methods for protein structure prediction based on homology modeling were developed starting in the 1990s and implemented in popular software suites, including the early version of Rosetta [Bowers et al (2000)] and others (e.g. SWISS-MODEL [Guex et al (1997)]). These methods heavily depend on statistical data. They involve aligning and ranking sequences and structures and parameterizing scoring functions through extensive analysis of sequence and structure databases. This process culminates in distilling the information into a few optimal structures or interaction models. [Wang et al (2019)]. Over the years, the growing volume of statistical data has necessitated the automation of tasks, particularly in searching and comparing information. Advancements in hardware architecture and storage capacity have supported this shift.Meanwhile, automatically trained neural networks (NN) have emerged as a natural solution for the "distillation" of this data [Kanada et al (2024)]. During the second decade of 2000s, the co-evolution of computer performance and algorithms led to the transition from machine learning (ML) to deep learning (DL). This shift involved adding layers to the neural networks, resulting in qualitative and quantitative predictive power improvements. The combination of an established supportive environment, the availability of big data, and the rise of DL has significantly contributed to the success of AI methods in bio-modeling.Specifically regarding protein structure, AlphaFold now achieves an impressive 99% accuracy in predicting single-chain proteins, rendering the CASP challenge-historically focused on structure prediction-less relevant. Besides the modeling of protein structures, a significant domain of artificial intelligence application elucidated by statistical analysis pertains to drug development. In particular, ML is used to address structure-activity relationships [Gupta et al (2021)] and uptake-toxicity of the drug [De Carlo et al (2024)], virtual screening, and structure-based design. While not claiming to cover all potential applications, we note that optimizing force fields for low-resolution models of biomolecules significantly benefits from machine learning [Kanada et al (2024), Majewski et al (2023), Mirarchi et al (2024)], whereas the application of graph neural networks for calculating molecular dynamical trajectories is a cutting-edge approach [Husik et al (2020)].","journal":"Frontiers in Molecular Biosciences","year":2025,"id":556981,"datarank":0.0,"base_score":0.0,"endowment":0.0,"self_citation_contribution":0.0,"citation_network_contribution":0.0,"self_endowment_contribution":0.0,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":1,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.9386,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2025-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":378845,"name":"Cecilia Giulivi","orcid":"0000-0003-1033-7435","position":1,"is_corresponding":false},{"id":1456658,"name":"Valentina Tozzini","orcid":"0000-0002-7586-5039","position":0,"is_corresponding":true}],"reference_count":17,"raw_metadata":null,"created_at":"2026-07-19T02:55:13.130091Z","pmid":"39830980","pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}