Artificial intelligence in the life sciences最新文献

{"title":"CDTA: Consistency-based deep learning method for drug-target affinity prediction using sequential features from protein language model","authors":"Jiffriya Mohamed Abdul Cader ,&nbsp;M.A. Hakim Newton ,&nbsp;Abdul Sattar","doi":"10.1016/j.ailsci.2026.100160","DOIUrl":"10.1016/j.ailsci.2026.100160","url":null,"abstract":"<div><div>Accurate drug-target affinity prediction is a critical phase for accelerating and reducing the failure rate of the drug discovery process. It is more challenging in the absence of reliable 3D structural information. Even though protein 3D structure prediction methods have improved, they still lack reliability and generalisability. Ensemble methods attempt to address these shortcomings, but these techniques themselves lack consistency in the values across different models. Therefore, we introduce a novel consistency-based deep learning method named CDTA using sequential features, which consists of a main prediction module and a supporting surrogate module, and the outputs of both modules are incorporated to improve the prediction accuracy. The method employs a novel feature combination, integrating explicit and deep learning-derived features from a protein language model in the prediction module. The CDTA achieves a Pearson correlation coefficient above 0.86 and a Concordance index (CI) over 0.87 on benchmark datasets KIBA, Davis, and BindingDB. Further, it surpasses the current state-of-the-art methods in the accuracy metric <span><math><msubsup><mrow><mi>r</mi></mrow><mrow><mi>m</mi></mrow><mrow><mn>2</mn></mrow></msubsup></math></span> across all three datasets. Lightweight and competent, CDTA does not rely on 3D structural information. Its simple architecture enables processing of larger datasets on limited hardware without compromising accuracy. Moreover, it facilitates not only the early identification of potential drug candidates but also helps to minimise the failure rate of drug discovery. Ultimately, it accelerates the process and reduces associated costs.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100160"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SynthFormer: Equivariant pharmacophore-based generation of synthesizable molecules for ligand-based drug design","authors":"Zygimantas Jocys ,&nbsp;Zhanxing Zhu ,&nbsp;Henriette M.G. Willems ,&nbsp;Katayoun Farrahi","doi":"10.1016/j.ailsci.2025.100148","DOIUrl":"10.1016/j.ailsci.2025.100148","url":null,"abstract":"<div><div>Drug discovery is a complex, resource-intensive process requiring significant time and cost to bring new medicines to patients. Many generative models aim to accelerate drug discovery, but few produce synthetically accessible molecules. Conversely, synthesis-focused models do not leverage the 3D information crucial for effective drug design. We introduce SynthFormer, a novel machine learning model that generates fully synthesizable molecules, structured as synthetic trees, by introducing both 3D information and pharmacophores as input. SynthFormer features a 3D equivariant graph neural network to encode pharmacophores, followed by a Transformer-based synthesis-aware decoding mechanism for constructing synthetic trees as a sequence of tokens. This provides capabilities for designing active molecules based on pharmacophores, exploring the local synthesizable chemical space around hit molecules and optimizing their properties. We demonstrate its effectiveness through various challenging tasks, including designing active compounds for a range of proteins, performing hit expansion and optimizing molecular properties.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100148"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145898134","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The comfort of automation: why cognitive sovereignty matters in AI-driven life sciences","authors":"Francesco Branda,&nbsp;Massimo Ciccozzi","doi":"10.1016/j.ailsci.2026.100158","DOIUrl":"10.1016/j.ailsci.2026.100158","url":null,"abstract":"<div><div>The integration of artificial intelligence (AI) into the life sciences is radically transforming research, clinical diagnosis, and therapeutic development processes, redefining the relationship between knowledge, decision-making, and responsibility. Advanced tools, from generative models to clinical assistants such as ChatGPT Health, offer greater efficiency, predictive power, and access to data, but carry significant risks of automation bias, epistemic delegation, and loss of professional skills. This article analyzes how the extensive use of AI can threaten cognitive sovereignty, i.e., the ability of researchers and professionals to critically evaluate and contextualize information generated by algorithms. It examines the emerging regulatory landscape, with a focus on the EU Artificial Intelligence Act, Food and Drug Administration (FDA) guidelines, European Medicines Agency (EMA) Good Machine Learning Practice (GMLP) principles, and World Health Organization (WHO) recommendations, which aim to ensure human oversight, transparency, and accountability. Technological tools and training approaches are discussed to mitigate risks such as silent errors, algorithmic dependence, and skill deterioration, promoting AI integration that reinforces human judgment without replacing it. The analysis highlights that the future of life sciences will depend not only on the technical capabilities of models, but also on the critical awareness with which they are used, focusing on training, governance, and responsible AI design.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100158"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189561","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A client-enhanced language model for interactive compound optimization guided by explainable artificial intelligence","authors":"Atsushi Yoshimori ,&nbsp;Jürgen Bajorath","doi":"10.1016/j.ailsci.2026.100154","DOIUrl":"10.1016/j.ailsci.2026.100154","url":null,"abstract":"<div><div>Compound optimization is of central relevance in medicinal chemistry. We introduce a new machine learning framework for iterative chemical optimization that integrates compound potency predictions, the explanation of predictions, and generative modeling and that is applicable to individual compounds. The approach identifies substituents in active compounds that limit their potency and iteratively replaces these substituents with others supporting potency increases. In proof-of-concept calculations, the methodology effectively optimizes compound potency. Furthermore, the optimization framework is combined with a large language model via the model concept protocol to generate an AI agent system for interactive optimization. The system is shown to successfully carry out optimization tasks of increasing complexity based on simple prompts, without the need for additional fine-tuning. The interactive computational optimization approach is accessible to non-experts and expected to be of particular interest for practical medicinal chemistry.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100154"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145980264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Implementing XAI in life sciences: Key challenges and pathways to solutions","authors":"Ondrej Krejcar ,&nbsp;Jamaluddin Abdullah ,&nbsp;Hamidreza Namazi","doi":"10.1016/j.ailsci.2026.100153","DOIUrl":"10.1016/j.ailsci.2026.100153","url":null,"abstract":"<div><div>The growing adoption of artificial intelligence (AI) in life sciences has been paralleled by growing concerns regarding transparency, interpretability, and trustworthiness of predictive models. While explainable artificial intelligence (XAI) has emerged as a powerful framework to bridge this gap, its practical deployment continues to face substantial technical, ethical, and regulatory barriers. This review provides a comprehensive overview of the challenges associated with implementing XAI in life science applications—including data complexity, model heterogeneity, computational costs, clinical integration, and ethical considerations—and discusses potential solutions and strategies to address them. By mapping recent advances in methodological approaches, regulatory frameworks, and interdisciplinary collaborations, we highlight a roadmap for embedding explainability into the AI lifecycle. The paper concludes with future perspectives on harmonizing interpretability with predictive performance in critical domains such as drug discovery, medical diagnostics, and bioinformatics.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100153"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145980263","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Long short-term memory-based chemical language models for bioactive molecular generation using tailored pre-training datasets","authors":"Ryuto Abe ,&nbsp;Tomoyuki Miyao","doi":"10.1016/j.ailsci.2026.100159","DOIUrl":"10.1016/j.ailsci.2026.100159","url":null,"abstract":"<div><div>Chemical language models (CLMs) computationally generate molecules as line notations, and their usefulness has been demonstrated in various applications. A common approach to producing target-specific molecules is to pre-train a CLM on a large molecular library and then fine-tune it on a dataset of targeted compounds. In this study, we systematically examine how pre-training datasets influence subsequent fine-tuning. For this purpose, six pre-training datasets were prepared, including those with and without structurally liable molecules estimated using RDKit functions, as well as bioactive and non-bioactive molecules assembled from publicly available databases. Six long short-term memory (LSTM)-based CLMs were pre-trained on the six pre-training datasets and then fine-tuned on five target-oriented molecular datasets. We revealed that selecting a pre-training dataset aligned with the researcher's design objective is key to controlling LSTM-based CLMs during fine-tuning, enabling the generation of non-liable and/or selective bioactive molecules without sacrificing diversity, similarity to the fine-tuning dataset, or test-molecule rediscovery.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100159"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189562","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The time dimension matters: Improving mode of action classification with live-cell imaging","authors":"Edvin Forsgren ,&nbsp;Jonne Rietdijk ,&nbsp;David Holmberg ,&nbsp;Julia Juneblad ,&nbsp;Bianca Migliori ,&nbsp;Martin M. Johansson ,&nbsp;Jordi Carreras-Puigvert ,&nbsp;Johan Trygg ,&nbsp;Gillian Lovell ,&nbsp;Ola Spjuth ,&nbsp;Pär Jonsson","doi":"10.1016/j.ailsci.2025.100152","DOIUrl":"10.1016/j.ailsci.2025.100152","url":null,"abstract":"<div><div>Morphological profiling is a common approach to investigate the modes of action (MOAs) of compounds. Most methods rely on fixed-cell assays, which provide only a single snapshot at a predefined time point and overlook the dynamic nature of cellular responses. In contrast, live-cell imaging tracks responses over time, offering deeper insight into compound-specific effects and mechanisms; however, time-series analysis of image data remains challenging due to limited analytical tools.</div><div>We present Live Cell Temporal Profiling (LCTP), a workflow for morphological profiling of label-free live-cell time series data that yields interpretable, biologically relevant results. We showcase LCTP in an MOA classification study using label-free data. The workflow integrates established deep-learning components, cell segmentation, live/dead classification, and single-cell feature extraction, with data-driven models to capture MOA-specific temporal phenotypes and produce time-resolved profiles that can be compared across compounds and cell lines.</div><div>We assess MOA classification performance using double-blinded cross-validation simulating a real-world screening scenario. LCTP significantly improves MOA classification over single–time point analysis, consistently across both cell lines used in the study. Time-resolved phenotypic modelling reveals transient, sustained, and delayed responses, clarifying compound-specific temporal effects and mechanisms across MOAs.</div><div>The presented workflow is modular: each step removes irrelevant information, enriching signal, and enabling straightforward updates as technologies evolve and as new technologies become available, while supporting reuse across studies broadly. We believe LCTP adds substantial value to high-throughput compound screening, showing that live-cell imaging combined with this workflow yields informative visualizations of temporal effects and improved MOA classification.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100152"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145980265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Drug sensitivity prediction across cancer cell lines with a focus on sarcoma","authors":"Elena Xerxa ,&nbsp;Selina Koch ,&nbsp;Silvia Vanni ,&nbsp;Alessandro De Vita ,&nbsp;Jürgen Bajorath","doi":"10.1016/j.ailsci.2026.100157","DOIUrl":"10.1016/j.ailsci.2026.100157","url":null,"abstract":"<div><div>Screening of drugs on cancer cell lines is a first step in the search for compounds leading to cancer cell death, providing the basis for follow-up investigations. Cell line-based drug sensitivity data can be used for predictive modeling, for example, in the context of drug repurposing. In this work, we report a machine learning framework for the systematic prediction of drug sensitivity based on transcriptomic data from cancer cell line screening. Three different categories of complementary classification models are introduced to predict cell line sensitivities to known drugs based on encoded gene expression profiles, predict activity of new drugs for given cell lines based on compound fingerprints, and predict responses of new cell lines to new drugs based on combined molecular representations and gene expression profiles. The models are found to have adequate predictive performance for cell-based data and are shown to be applicable to predict drug sensitivity for sarcoma cell lines, a rare form of cancer, for which data are limited. For the subset of sarcoma cell lines, tested drugs were also ranked based on cell line sensitivity rates. Taken together, our results suggest that the complementary machine learning models have potential for practical applications to search for new compounds for the treatment of different types of cancers including sarcoma.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100157"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189565","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CLOVER: A framework for benchmarking synthetic data generation methods balancing utility and privacy in healthcare","authors":"Yue Qi ,&nbsp;Lorrie Herbault ,&nbsp;Hadrien Lautraite ,&nbsp;Michael Yu ,&nbsp;Katleen Blanchet ,&nbsp;Christian Vincelette ,&nbsp;Louis Mullie ,&nbsp;Guillaume Dumas ,&nbsp;Jean-François Rajotte ,&nbsp;Kamran Afzali ,&nbsp;Sébastien Gambs ,&nbsp;Michaël Chassé","doi":"10.1016/j.ailsci.2026.100155","DOIUrl":"10.1016/j.ailsci.2026.100155","url":null,"abstract":"<div><h3>Background</h3><div>Synthetic data enables open and efficient medical research by enhancing real-world data. We examined the utility-privacy trade-off of synthetic data generated with and without differential privacy (DP) using CLOVER, a novel open-source Python library that we have developed.</div></div><div><h3>Methods</h3><div>We generated synthetic datasets based on data from MIMIC-III (24 variables, n = 15,118) and eICU (23 variables, n = 3,726). The generative approaches used were SMOTE, DataSynthesizer, Synthpop, MST, CTGAN, TVAE, CTABGAN+, and FinDiff, with and without DP. We evaluated the utility and privacy of the generated datasets based on univariate, bivariate, and population fidelity; analysis-specific and distance-based metrics; and membership inference attacks (MIAs). We benchmarked the synthetic datasets using rank-derived scores for utility and privacy. We examined the impact of DP on machine learning (ML) performance and MIAs and analyzed the achievable utility-privacy trade-off by generating synthetic data across a range of privacy regimes. We compared computational resource usage across generators.</div></div><div><h3>Findings</h3><div>When fully relaxing DP constraints, MST (<em>ε</em> = 10⁵ and <em>δ</em> = 0·9999) ranked the most private on MIMIC-III and the second most private on eICU (DCR and NNDR well above baseline for both datasets, top 1% precision for MIAs below 0·53 and 0·62 for MIMIC-III and eICU, respectively) but placed 7th out of eight in utility for both datasets. Conversely, Synthpop ranked first in utility for both datasets. It achieved Hellinger distance of 0·88 × 10^-2 and 1·41 × 10^-2; pairwise correlation difference of 0·31 and 0·68; distinguishability of 0·02 × 10^-1 and 0·02 × 10^-1; AUC difference of 0·20 × 10^-1 and 0·10 × 10^-1 on classification task; and RMSE difference of 2·53 × 10^-2 and 13·64 × 10^-2 on regression tasks for MIMIC-III and eICU, respectively. However, it ranked 7th in privacy for both datasets. SMOTE and TVAE were each outperformed by at least one other generator in terms of both utility and privacy based on the rank-derived scores for both datasets. Once DP was introduced, utility decreased across all algorithms, with no method consistently outperforming others across all privacy regimes.</div></div><div><h3>Interpretation</h3><div>There is a tradeoff between utility and privacy in the non-DP setting. DP reduced utility but ensured a consistent level of privacy, allowing for a fair comparison of the utility of different generators. Selecting an appropriate generator depends on the privacy needs, intended use case and the user’s available resources.</div></div><div><h3>Funding</h3><div>Institute for Data Valorisation (IVADO - PRF-2021-03).</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100155"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146189564","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a deep neural network model for simultaneous analysis of extracellular analyte gradients for a population of cells","authors":"Ivon Acosta-Ramirez ,&nbsp;Ferhat Sadak ,&nbsp;Sruti Das Choudhury ,&nbsp;James Thomson ,&nbsp;Salome Perez-Rosero ,&nbsp;Portia N.A. Plange ,&nbsp;Sofia E. Morales-Mendivelso ,&nbsp;Nicole M. Iverson","doi":"10.1016/j.ailsci.2026.100156","DOIUrl":"10.1016/j.ailsci.2026.100156","url":null,"abstract":"<div><div>Detecting the spatial release of extracellular nitric oxide (NO) is essential for understanding the dynamics in cell communication for physiological and pathological processes. This study presents an innovative methodology that integrates fluorescence-based sensing platforms utilizing single walled carbon nanotubes (SWNT) with machine learning models to expedite the spatial data analysis of extracellular analytes. The deep learning model You Only Look Once (YOLOv8) segmentation achieves accurate cell identification across diverse morphologies and clustered cell groups, with a recall of 98% and a precision of 83%. The spatial analysis of extracellular NO is achieved by extracting the cell contour coordinates from the YOLO-identified cells and translocating the boundaries onto SWNT fluorescence files. The model enables rapid analysis for multiple cells across numerous images, with 100 image pairs completed in just 68 s. The combination of nanotechnology with automated neural network-based cell detection establishes a robust sensing framework with pixel-level spatial resolution of NO dynamics, delivering critical insights into cellular communication and holding promising implications for diagnostic and therapeutic applications.</div></div>","PeriodicalId":72304,"journal":{"name":"Artificial intelligence in the life sciences","volume":"9 ","pages":"Article 100156"},"PeriodicalIF":5.4,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}