OpenAI's GPT‑Rosalind model, designed for chemistry and genomics, demonstrates superior performance on scientific benchmarks, outperforms human experts, offers a rich plugin ecosystem, and implements strict access controls to help accelerate early-stage drug research while ensuring responsible AI use in life sciences.
The scLong foundation model, trained on 48 million cells and 28 k genes, integrates full‑gene expression with Gene Ontology knowledge to outperform existing methods on genetic perturbation, chemical response, cancer drug prediction, gene‑regulatory network inference, and batch integration tasks.
MIT and ETH Zurich introduce APOLLO, a deep‑learning autoencoder that learns a partially overlapping latent space to explicitly disentangle shared and modality‑specific information in multimodal single‑cell datasets, demonstrating superior cell‑type classification, cross‑modal prediction, and protein localization insights across sequencing and imaging data.
Google DeepMind’s AlphaGenome, featured on Nature’s cover, reads up to one million DNA bases at once, predicts the functional impact of any mutation across gene expression, splicing, chromatin and protein binding, and outperforms prior models by more than double on key benchmarks.
Researchers present LucaOne, a Transformer‑based foundation model that unifies DNA/RNA and protein sequences using a 39‑token vocabulary, rotary positional encoding, and molecule‑type embeddings, and demonstrate through extensive multi‑task benchmarks that it outperforms domain‑specific models across seven biological tasks.
RFdiffusion2 introduces a novel deep generative approach that eliminates residue enumeration and sequence indexing, enabling atom‑level protein backbone generation from simple chemical reaction descriptions, achieving a 100% success rate across 41 benchmark cases and providing a step‑by‑step demo on the OpenBayes platform.
scvi‑hub, introduced by UC Berkeley researchers, provides a model‑driven platform that compresses, versions, and shares large single‑cell genomics datasets via pretrained probabilistic models, enabling fast, reproducible analysis and broad community reuse while addressing data‑size and training bottlenecks.
Researchers at the Arc Institute and Stanford unveiled Evo 2, an AI model capable of designing entire viral genomes, and demonstrated its power by generating the first AI‑crafted ΦX174 phage genome, detailing the annotation pipeline, fine‑tuning, validation, and evolutionary insights.
BackFlip is an equivariant neural network that predicts per‑residue flexibility from protein backbone structures, while the derived FliPS conditional flow‑matching model generates novel backbones matching desired flexibility distributions, demonstrated through diverse designs validated by molecular dynamics simulations, with code available on GitHub.
Microsoft Research’s AI for Science team released the open‑source BioEmu model, a generative diffusion architecture that leverages AlphaFold’s Evoformer and extensive MD and stability data to efficiently sample protein conformational ensembles, achieving near‑MD accuracy in free‑energy and mutation stability predictions while dramatically reducing computational cost.
The All‑Atom Protein Generative Model (APM) introduced by Hunan University, UCAS and ByteDance Seed combines full‑atom representation, multi‑chain native modeling, and novel flow‑matching techniques to outperform existing SOTA methods on folding, reverse‑folding, antibody and peptide design tasks, backed by a curated multi‑source dataset and extensive benchmarks.
The SUICA model combines graph autoencoders and implicit neural representations to denoise, impute, and super‑resolve spatial transcriptomics data, overcoming resolution‑cost trade‑offs and dropout noise, and delivers biologically richer gene expression predictions validated on mouse embryo and brain datasets.
DeepPurpose, a PyTorch‑based AI toolkit developed by Harvard researchers, provides COVID‑19 bioassay data and 56 cutting‑edge models that enable rapid drug‑target affinity prediction, virtual screening, and drug repurposing with just a few lines of code, dramatically shortening new‑drug development cycles.
This article reviews the evolution from AlphaFold 2 to AlphaFold 3, introduces Baidu's HelixFold 3 as the first domestic model matching AlphaFold 3, compares their benchmark results on small‑molecule ligands, nucleic acids and protein complexes, and explains the cloud‑based service and confidence scoring that make high‑throughput structure prediction accessible.
The inaugural Shanghai International Computational Biology Innovation Competition concluded with five finalist teams, awarding a GeminiMol team from ShanghaiTech University first prize for discovering highly active molecules validated by wet experiments, while the event showcased AI-driven drug discovery targeting the NMDA receptor, highlighted the city’s new computational biology action plan, and emphasized talent selection through competition.
This article explains how to use the Kubernetes-native Argo Workflows engine to automate and scale complex gene-computing pipelines, detailing its advantages, challenges, and a step-by-step BWA alignment workflow example on Alibaba Cloud’s ACK platform.
HelixFold‑Multimer, a new Baidu PaddleHelix model, outperforms AlphaFold 3 on antigen‑antibody and peptide‑protein complex predictions, achieving mean DockQ scores of 0.41 and 0.38 respectively and success rates up to 77 % when epitope data are used, and is already deployed in large‑molecule drug pipelines.
This article reviews the fundamentals of graph neural networks and graph representation learning, explains why proteins can be modeled as graphs, and surveys recent GNN‑based applications such as structure prediction, function annotation, protein design, and self‑supervised representation learning, concluding with future research directions.
The article introduces the MEGA-Protein tool, which integrates an AI‑driven MSA engine and the MindSpore framework to overcome AlphaFold 2’s limitations and achieve high‑accuracy protein structure prediction, and also announces the MSG Enterprise Tour and Hangzhou Developer Day event on April 27.
This article reviews global trends in AI‑driven biopharma, outlines the technical challenges, and details Baidu Intelligent Cloud’s bio‑computing large‑model technologies—including HelixGEM, HelixFold, and HelixFold‑Single—along with their industrial applications in drug design, protein prediction, and mRNA vaccine development.
Virus phylogeny explores how genetic mutations, recombination, and evolutionary rates shape viral lineages, using tree-based methods such as distance, maximum parsimony, maximum likelihood, and Bayesian approaches, while addressing challenges like variable molecular clocks, recombination, and limited sampling to infer relationships and origins.
DeepMind’s AlphaFold inspired Uni‑Fold, now accelerated with Alibaba Cloud’s PAI platform, can predict protein structures up to 6.6k amino acids—covering 99.992% of known sequences—delivering ten‑minute inference for SARS‑CoV‑2 spike trimers and setting new performance benchmarks for AI‑driven structural biology.
This presentation by Stanford PhD student Huang Kexin explores the challenges and innovations of applying graph machine learning to molecular and biomedical networks, introducing specialized GNN architectures, actionable hypothesis generation, domain‑scientist interfaces, few‑shot learning, and the Therapeutics Data Commons for accelerating drug discovery.
This article presents the Adaptive Universal Generalized PageRank Graph Neural Network (GPR‑GNN), explains the two main limitations of existing GNNs—lack of generality across homophilic and heterophilic graphs and the over‑smoothing problem—and demonstrates through synthetic and real‑world experiments that GPR‑GNN achieves robust node classification while remaining interpretable and parameter‑efficient.
This guide explains how to efficiently locate executables, files, and content on Linux using commands like whereis, which, locate, find, and grep, with practical examples for filtering by name, time, size, depth, and pattern matching.
Alibaba’s DAMO Academy and Zhejiang CDC have built an AI‑powered, fully automated whole‑genome sequencing platform that reduces COVID‑19 sample analysis from several hours to about half an hour, delivering rapid, accurate detection of viral mutations and supporting vaccine and drug research.
Yang Zhentao’s 2018 conference talk surveys data‑visualization fundamentals, highlights the multidisciplinary skills required, introduces the open‑source Circos tool and its polar‑coordinate workflow, showcases genomic and business use cases, and compares alternative platforms, emphasizing data quality, query capability, and proper view selection.
