Fitness translocation: improving variant effect prediction with biologically-grounded data augmentation DOI Creative Commons
Adrien Mialland, Shusei Fukunaga,

Riku Katsuki

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Дек. 20, 2024

Abstract Navigating the protein fitness landscape is critical for understanding sequence-function relationships and improving variant effect prediction. However, limited availability of experimentally measured functional data poses a significant bottleneck. To address this, we present novel augmentation strategy called translocation, which leverages landscapes from related proteins to enhance performance predictors on target protein. Using embeddings language models by translocating features within sequence space, transfer information homologous datasets augment its dataset. Our approach was evaluated across diverse species, including IGPS orthologs, GFP SARS-CoV-2 spike strains cell entry ACE2 binding. The results demonstrate consistent substantial improvements in predictive performances, particularly with training data. Furthermore, introduce systematic selection framework identifying most beneficial optimizing gains. This study highlights potential translocation advance engineering implementation method available at https://github.com/adrienmialland/ProtFitTrans .

Язык: Английский

Influence of next-generation artificial intelligence on headache research, diagnosis and treatment: the junior editorial board members’ vision – part 2 DOI Creative Commons
Igor Petrušić, Chia‐Chun Chiang, David García‐Azorín

и другие.

The Journal of Headache and Pain, Год журнала: 2025, Номер 26(1)

Опубликована: Янв. 2, 2025

Part 2 explores the transformative potential of artificial intelligence (AI) in addressing complexities headache disorders through innovative approaches, including digital twin models, wearable healthcare technologies and biosensors, AI-driven drug discovery. Digital twins, as dynamic representations patients, offer opportunities for personalized management by integrating diverse datasets such neuroimaging, multiomics, sensor data to advance research, optimize treatment, enable virtual trials. In addition, devices equipped with next-generation biosensors combined multi-agent chatbots could real-time physiological biochemical monitoring, diagnosing, facilitating early attack forecasting prevention, disease tracking, interventions. Furthermore, advances discovery leverage machine learning generative AI accelerate identification novel therapeutic targets treatment strategies migraine other disorders. Despite these advances, challenges standardization, model explainability, ethical considerations remain pivotal. Collaborative efforts between clinicians, biomedical biotechnological engineers, scientists, legal representatives bioethics experts are essential overcoming barriers unlocking AI's full transforming research healthcare. This is a call action proposing frameworks AI-based into care.

Язык: Английский

Процитировано

2
Artificial Intelligence in Precision Medicine and Patient-Specific Drug Design DOI Open Access

Shudhanshu Ranjan,

Arpita Singh,

Ruchi Yadav

и другие.

Biomedical & Pharmacology Journal, Год журнала: 2025, Номер 18(December Spl Edition), С. 283 - 294

Опубликована: Янв. 20, 2025

Artificial intelligence (AI) has emerged as a transformative force in personalized healthcare and precision medicine over the past decade. AI techniques like machine learning, deep natural language processing make possible study of huge quantities heterogeneous patient records from electronic health records, genomic profiles, wearable devices, clinical trials. This allows for more accurate disease prediction, treatment planning, tailored drug discovery. Key areas impact include AI-driven biomarker discovery, virtual screening, de novo design, pharmacogenomics. The integration is revolutionizing multiple aspects medicine, identifying novel therapeutic targets to optimizing trial design dosing. algorithms can detect subtle patterns complex biological data, predict drug-target interactions, simulate molecular behaviour accelerate typically costly time-consuming development process. However, challenges remain around data quality, privacy, algorithmic bias, equitable implementation. Ethical considerations regarding genetic discrimination informed consent also need be carefully addressed. review examines current applications, challenges, future directions advancing patient-specific therapies development.

Язык: Английский

Процитировано

1
Multi-Criteria Decision Analysis in Drug Discovery DOI Creative Commons
Rafał A. Bachorz,

Michael S. Lawless,

David W. Miller

и другие.

Applied Biosciences, Год журнала: 2025, Номер 4(1), С. 2 - 2

Опубликована: Янв. 6, 2025

Drug discovery is inherently a multi-criteria optimization problem. In the first instance, it involves tremendously large chemical space, where each compound can be characterized by multiple molecular and biological properties. Modern computational approaches try to efficiently explore space in search of molecules with desired combination For example, Pareto optimizers identify so-called “Pareto front”, set non-dominated solutions. From qualitative perspective, all solutions on front are potentially equally desirable, expressing trade-off between goals. However, often there need weight objectives differently, depending their perceived importance. To address this, we recently implemented new Multi-Criteria Decision Analysis (MCDA) method as part AI-powered Design (AIDDTM) technology initiative. This allows user various objective functions which, turn, directs generative chemistry process toward areas space.

Язык: Английский

Процитировано

0
Fine-Tuning a Genetic Algorithm for CAMD: A Screening-Guided Warm Start DOI
Yifan Wang, Lorenz Fleitmann, Lukas Raßpe-Lange

и другие.

Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown

Опубликована: Фев. 18, 2025

More sustainable chemical processes require the selection of suitable molecules, which can be supported by computer-aided molecular design (CAMD). CAMD often generates and evaluates structures using genetic algorithms. However, algorithms suffer from slow convergence, might yield suboptimal solutions. In response to these challenges, this work presents a method fine-tune algorithm for CAMD. The proposed builds on COSMO-CAMD framework that utilizes solving optimization-based problems COSMO-RS predicting physical properties molecules. key idea is integrate results fast large-scale screening into through an automated fragmentation procedure. By generating promising initial population constructing tailored fragment library, our enables targeted initialization algorithm, referred as warm-start. applied in two case studies solvents extracting γ-valerolactone phenol, respectively, aqueous Compared benchmark method, warm-started achieves 70% faster discovers 4-fold more top-performing candidate identifies seven fragments, culminating discovery novel specifically phenol case. optimal solvent found all computational runs. Overall, significantly improves efficiency, effectiveness, robustness design.

Язык: Английский

Процитировано

0
Advancing drug discovery and development through GPT models: A review on challenges, innovations and future prospects DOI Creative Commons
Zhinya Kawa Othman, Mohamed Mustaf Ahmed, Olalekan John Okesanya

и другие.

Intelligence-Based Medicine, Год журнала: 2025, Номер 11, С. 100233 - 100233

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

0
GL4SDA: Predicting snoRNA-Disease Associations Using GNNs and LLM Embeddings DOI Creative Commons

Massimo La Rosa,

Antonino Fiannaca, Isabella Mendolia

и другие.

Computational and Structural Biotechnology Journal, Год журнала: 2025, Номер 27, С. 1023 - 1033

Опубликована: Янв. 1, 2025

Small nucleolar RNAs (snoRNAs) play essential roles in various cellular processes, and their associations with diseases are increasingly recognized. Identifying these snoRNA-disease relationships is critical for advancing our understanding of functional potential therapeutic implications. This work presents a novel approach, called GL4SDA, to predict using Graph Neural Networks (GNN) Large Language Models. Our methodology leverages the unique strengths heterogeneous graph structures model complex biological interactions. Differently from existing methods, we define set features able capture deeper information content related inner attributes both snoRNAs design GNN based on highly performing layers, which can maximize results this representation. We consider snoRNA secondary disease embeddings derived large language models obtain node features, respectively. By combining structural rich semantic diseases, construct feature-rich representation that improves predictive performance model. evaluate approach different architectures exploit capabilities many convolutional layers compare three other state-of-the-art graph-based predictors. GL4SDA demonstrates improved scores link prediction tasks its implication as tool exploring relationships. also validate findings through case studies about cancer highlighting practical application method real-world scenarios obtaining most important explainable artificial intelligence methods.

Язык: Английский

Процитировано

0
Limeade: Let integer molecular encoding aid DOI
Shiqiang Zhang, Christian Feldmann, Frederik Sandfort

и другие.

Computers & Chemical Engineering, Год журнала: 2025, Номер unknown, С. 109115 - 109115

Опубликована: Апрель 1, 2025

Процитировано

0
A review on computational tools for antidiabetic herbs research DOI Creative Commons
Sangeeta Sanjay Jadhav, Gargi Nikhil Vaidya, Amisha Vora

и другие.

Discover Chemistry., Год журнала: 2025, Номер 2(1)

Опубликована: Апрель 15, 2025

Язык: Английский

Процитировано

0
Artificial intelligence in drug discovery and development: transforming challenges into opportunities DOI Creative Commons
Shashi Kant,

Deepika Deepika,

S. C. Roy

и другие.

Опубликована: Июнь 2, 2025

Язык: Английский

Процитировано

0
Utilizing Omics Technologies and Machine Learning to Improve Predictive Toxicology DOI Open Access

Ahrum Son,

Jongham Park, Woojin Kim

и другие.

Опубликована: Авг. 26, 2024

The topic of predictive toxicology has been greatly influenced by recent progress in comprehending drug toxicity processes and enhancing medication development. integration omics technologies, such as transcriptomics, proteomics, metabolomics, with traditional toxicological assessments yielded extensive knowledge about the biological pathways implicated drug-induced toxicity. utilization a multi-omics method amplifies ability to identify biomarkers that can detect at an early stage, hence safety profile novel therapeutic medicines. Machine learning silico models, QSAR models multi-task deep algorithms, have become essential tools. They shown great accuracy predicting endpoints helped identification new targets. introduction microphysiological systems PBPK modeling enhanced transfer preclinical discoveries clinical results, providing more precise forecasts human reactions medications. Notwithstanding these progressions, obstacles diversity data complex nature require sophisticated computational techniques for efficient analysis. Continued cooperation established procedures are crucial fully utilize guaranteeing creation safer medicinal agents.

Язык: Английский

Процитировано

2