Publications by authors named "K G Leidal"
Understanding biomolecular interactions is fundamental to advancing fields like drug discovery and protein design. In this paper, we introduce Boltz-1, an open-source deep learning model incorporating innovations in model architecture, speed optimization, and data processing achieving AlphaFold3-level accuracy in predicting the 3D structures of biomolecular complexes. Boltz-1 demonstrates a performance on-par with state-of-the-art commercial models on a range of diverse benchmarks, setting a new benchmark for commercially accessible tools in structural biology.
View Article and Find Full Text PDFArticle Synopsis
- - Clinical trials for metabolic dysfunction-associated steatohepatitis (MASH) need accurate histologic scoring to assess participants and outcomes, but varying interpretations have affected results.
- - The AI-based tool AIM-MASH showed strong consistency and agreement with expert pathologists in scoring MASH histology, achieving accuracy comparable to that of average pathologists.
- - AIM-MASH demonstrated a strong ability to predict patient outcomes, correlating well with pathologist scores and noninvasive biomarkers, indicating it could enhance the efficiency and reliability of clinical trials for MASH.
Article Synopsis
- The study focuses on creating a deep learning digital pathology tool for accurately detecting, segmenting, and classifying nuclei in cancer tissues, addressing challenges in quantifying nuclear morphology in histologic images.
- This tool was trained on nucleus annotations to analyze H&E-stained slides from various cancer cohorts (BRCA, LUAD, PRAD), revealing significant differences in nuclear features like shape and size linked to genomic instability and cancer prognosis.
- Results highlighted that certain nuclear characteristics, particularly in fibroblasts, were associated with patient survival outcomes and gene expression related to tumor biology, paving the way for better understanding of cancer biomarkers.
Article Synopsis
- Clinical trials for nonalcoholic steatohepatitis (NASH) rely on consistent histologic scoring, but variability in these interpretations has affected trial results.* -
- An AI tool called AIM-NASH was developed to provide standardized scoring for NASH histology, showing strong correlation with expert consensus scores and improving predictive accuracy for patient outcomes.* -
- In a retrospective analysis, AIM-NASH helped meet previously unmet pathological endpoints in the ATLAS trial, suggesting it could reduce variability in scoring and enhance the assessment of treatment responses in clinical trials.*
Polymorphonuclear leukocytes (PMN) phagocytose and kill individual bacteria but are far less efficient when challenged with bacterial aggregates. Consequently, growth within a biofilm affords Staphylococcus aureus some protection but PMN penetrate S. aureus biofilms and phagocytose bacteria, suggesting that enzymes released through neutrophil degranulation degrade biofilms into fragments small enough for phagocytosis.
View Article and Find Full Text PDF