What happened
Breast cancer remains one of the leading causes of cancer-related deaths among women worldwide. Drug resistance, toxicity, and limited target specificity are the major challenges in the development of effective therapeutics. Anti-breast cancer peptides (ABCPs) have emerged as effective drug candidate due to its low toxicity, high selectivity, and ability to target cancer. However, it is very time consuming and expensive to identify novel ABCPs only through experiment methods. To address this challenge, we developed ABCP_finder, the first dedicated computational framework specifically designed for the prediction of ABCPs using transformer-based protein language model embeddings. Positive and negative datasets were carefully constructed to ensure a biologically meaningful classification task. To prevent the data leakage and realistic evaluation, homology aware train-test split strategy was utilized by using CD-HIT at 30% sequence identity with 80% coverage. Peptide representations were generated using pretrained transformer models, ProtBERT and ESM2, followed by classification using a multilayer perceptron (MLP). Among the tested models, ProtBERT showed superior performance, achieving 93.82% accuracy, 86.88% recall, 90.59% F1-score, 0.8618 MCC, 96.67% AUC, and a Brier Score of 0.0633, demonstrating strong predictive capability under imbalanced conditions. Calibration analysis supported the selection of a 0.7 probability threshold for identifying high confidence ABCPs. Further external validation using xDeep-AcPEP demonstrated that unknown peptide sequences that has been predicted as ABCPs by ABCP_finder are exhibiting favourable IC values. This supports the biological relevance of these unknown peptides. Overall, ABCP_finder provides a reliable and practical platform for large-scale ABCP screening and can significantly accelerate the discovery of novel peptide therapeutics for breast cancer treatment. Update (2026-09-15): Suspected sabotage causes major Netherlands rail disruption. Article URL: https://www.bbc.com/news/articles/c8ly49w9g1edo Comments URL: https://news.ycombinator.com/item?id=49710253 Points: 146 # Comments: 116 Update (2026-09-16): ML4SD: Leveraging Machine Learning and High-Throughput Search Algorithms for an Iterative Growth-Coupled Design Innovation . Optimizing microbial biomanufacturing is required if renewable and waste carbon are to replace petrochemical routes at competitive titers, rates, and yields. Growth-coupled (GC) production supports that goal by linking target synthesis to biomass formation, so product formation is required for growth. Constructing knockout strains yielding GC production from a list of candidate genes is labor and time demanding. This results in few in vivo tested designs, which hampers standard machine-learning methods to learn GC patterns for specific bioprocesses. We therefore developed ML4SD, an active-learning Design-Build-Test-Learn (DBTL) cycle that trains ensembles on genome-scale metabolic model (GEM) scores of knockout designs, sampling the next designs from predicted model performance and error. That cycle generalizes only if the initial library is large and diverse, including suboptimal and non-viable designs; libraries restricted to minimal designs or Pareto-optimal knockouts were found to generate models overfitting. To meet those speci Update (2026-09-18): Physical priors improve performance of structure-based binding affinity models . Structure-based drug discovery is a widely used paradigm for the rational design of novel small molecule therapeutics. However, the benefits conferred by the use of structural information has seen limited adoption in machine learning, where ligand-only ("2D") models are still the industry standard for molecular property or binding affinity prediction. Structure-based ("3D") ML models for binding-affinity prediction promise to present a clear advantage, but have not yet overtaken existing 2D models. Here, we show that physics-based priors can improve predictive performance of structure-based models by comparing different model architectures with varying physical priors on several prediction tasks. We present the Modular Training and Evaluation of Neural Networks (mtenn) package, where we decompose affinity prediction into separate steps of embedding structure into learned representations and combining those embeddings into a predicted binding affinity. We consider both E(3)-invariant and E(3)-equivariant architectures to determine the importance of encoding roto-translational Update (2026-09-18): A Unified 3D Generative Model for Synthesizable Structure-Based Drug Design . Traditional screening-based drug discovery is inherently limited by the astronomical scale of the chemical space. Generative modelling offers a compelling alternative to the classical search paradigm and enables rational, bottom-up design of novel and target-specific small molecules. However, its impact has been hampered by challenges in synthetic accessibility of the designed compounds and lack of large-scale experimental validation. Here, we introduce LDDM (Large Drug Discovery Model), a generative framework that supports a range of drug discovery tasks, including constrained and unconstrained docking, fragment linking and growing, and de novo design. We further introduce a programmable design algorithm that enables accurate design of synthetically accessible compounds satisfying various fine-grained objectives. We experimentally validated the designed or optimised ligands for five therapeutically relevant protein targets. In all cases, LDDM achieved high success rates, allowing us to identify molecules with confirmed binding affinity while synthesizing only a small number of