Tag
SimpleDesign introduces a joint model for protein sequence and structure codesign, trained end-to-end in data space using a single-stage objective, achieving competitive performance on co-design and generation benchmarks.
TxBench is a benchmark for antibody discovery, likely assessing computational or AI-driven methods in this biotech domain.
Jingtian Zhou builds single-cell and computational frameworks to probe the 3D genome, contributing to major publications in Science, Cell, and Cancer Cell that advance understanding of cell states and diseases.
This paper introduces BixBench3, a benchmark for evaluating AI agents on computational biology tasks, revealing that frontier LLMs can reproduce approximately 48% of real research workflows but struggle with large datasets and sequential steps.
The paper identifies a failure mode called off-manifold collapse in guided protein language models, where generated sequences degenerate, and proposes Mahalanobis filtering as a training-free post-hoc method to improve sequence quality and structural plausibility.
GenEx is a novel graph-based pipeline that converts SARS-CoV-2 gene sequences into codon co-occurrence graphs to detect variants, using techniques like MSCG and LAPCG, and demonstrates effectiveness with benchmarked ML models.
The article explores whether a computer scientist could design a genome-sized program to build a brain from a single cell, demonstrating that algorithmic scaling constraints align with biological strategies, suggesting brain design is grounded in computational necessity.
Researchers created VirTues, a unified foundation model for spatial proteomics that translates diverse tissue imaging data into a standard language, enabling faster and more accurate medical diagnoses and personalized treatments.
A Nature research paper presents a rational design approach for intrinsically disordered proteins to investigate sequence–function relationships, introducing the open-source tool GOOSE and providing analysis scripts and data.
Chamaileon introduces a framework for multi-target and multi-state protein binder design using contextualized sequence-structure co-modeling and mixed sampling, achieving adaptability across diverse conformational landscapes and multi-target requirements.
Anthropic's Claude Opus 5 achieves state-of-the-art results across computational biology, protein modeling, and wet-lab reasoning, positioning it as a general-purpose AI biology agent.
A Reddit post summarizing a survey paper that covers 25 deep learning methods for scRNA-seq analysis, organized into 6 subcategories with a detailed table.
The Arc Institute hosted its 3rd Annual Investigator Retreat, featuring talks on genome engineering, generative protein models, computational biology, and deep dives into neurodegeneration, immunology, and metabolism.
Michael Antonov, co-founder of Oculus, discusses his pivot to drug discovery through Deep Origin, emphasizing the integration of AI with physics-based simulations and experimental validation for safer therapies, and the role of VR in medical training.
SpaCellAgent is a self-evolving LLM-based multi-agent framework that automates end-to-end spatiotemporal trajectory analysis from natural language queries, achieving over 40% improvement in analytical efficiency while maintaining expert-level performance.
NVIDIA BioNeMo Recipes provide step-by-step, parameter-efficient fine-tuning recipes for biological foundation models using LoRA, demonstrated on two case studies with ESM2-3B and Evo2-1B on a single workstation GPU.
CryoACE is an atom-centric framework for automated model building from cryo-EM density maps, featuring iterative refinement and training-free guidance using local resolution priors. It significantly outperforms existing methods on static benchmarks and reveals atomic-level dynamic conformations on complex datasets.
OpenAI introduces GeneBench-Pro, a research-level benchmark to test AI agents' ability to navigate messy biological data, choose analysis paths, and make judgment calls in computational biology.
Anthropic launches Claude Science, a desktop app for macOS and Linux that provides a unified research environment for life sciences, integrating AI, databases, HPC, and tools for genomics, proteomics, structural biology, and more.
OpenAI introduces GeneBench-Pro, a research-level benchmark designed to test AI agents' ability to perform judgment-heavy analyses in computational biology, covering genomics, quantitative biology, and translational medicine.