螺旋桨PaddleHelix

A high-performance biological computation platform, leveraging the Wenxin foundation model for science and Baidu Cloud's high-performance computing cluster, is designed to support emerging industries such as biopharmaceuticals, biological breeding, vaccine design, and biomaterials.

HelixFold3 Update Zone

Antibody Design Beta Launch
Utilize HelixFold3 to implement high-active small protein binding screening, and the experimental success rate is high
New
HF3 Inference Price Drop
With HelixFold3's significant performance boost, inference costs are nearly 70% off
New
General API
HelixFold3 has supported large-scale protein calling via API and the results are available for commercial use

Latest News

Molecular Design Update
AI-powered scaffold-based molecule generation, significantly improved synthetic accessibility.
New
Industry Enabler
Advance AI-driven peptide drug research with PepLib
User Spotlight
Discover how developers use PaddleHelix to innovate anti-cancer peptide drugs.

<h2 class="jsx-1857088743 jsx-3833697322">HelixFold3 Foundation Model for Any Molecular Interactions</h2><p class="jsx-1857088743 jsx-3833697322 learn-more">Learn More<span role="img" aria-label="right" class="anticon anticon-right"><svg viewBox="64 64 896 896" focusable="false" data-icon="right" width="1em" height="1em" fill="currentColor" aria-hidden="true"><path d="M765.7 486.8L314.9 134.7A7.97 7.97 0 00302 141v77.3c0 4.9 2.3 9.6 6.1 12.6l360 281.1-360 281.1c-3.9 3-6.1 7.7-6.1 12.6V883c0 6.7 7.7 10.4 12.9 6.3l450.8-352.1a31.96 31.96 0 000-50.4z"></path></svg></span></p>

HelixFold3 is a model that supports the interaction prediction of common small molecules, proteins, DNA, RNA and ions.

Application Scenarios

Small Protein Design
HelixFold3 enables efficient design of specific binders, generating high-success-rate candidates with just the target sequence, leading to higher experimental success.
Antibody Design
HelixFold3's high-precision antigen-antibody structural predictions enable the design of novel, high-potential antibody molecules, enhancing the efficiency of antibody drug development by optimizing antibody-antigen interactions.
Enzyme Protein Design
By predicting structures, HelixFold3 designs highly active enzymes, speeding up the optimization of natural enzymes and providing innovative solutions for green chemistry and biocatalysis, while reducing ineffective candidates.

Calling Methods

Bio-computing Platform
Login and use it all at once, submit process and results are visualized, greatly improving operation efficiency
Use Now
API Calling
Supports the submission of large batches of tasks in your own workflow, suitable for downstream screening and design
View Description

Versatile solutions for various scenarios

Small Molecule Drug Discovery

Peptide Drug/Biologics Discovery

mRNA Therapeutics Discovery

Plant Breeding, Enzyme Engineering, Gene Therapy

Coming Soon

Small Molecule Drug Discovery

At pivotal stages such as hit compound exploration, lead compound optimization, candidate molecule selection, and patent application, our comprehensive capabilities, spanning foundation models, drug-likeness prediction, molecular conformation estimation, high-throughput virtual screening, as well as small molecule generation and optimization, empower our clients with highly effective hit discovery solutions and cost-efficient hit-to-lead programs. This rapid process enables them to swiftly identify optimal potential drug candidates within a condensed timeframe.

Hits Discovery

High-precision, high-performance conformation prediction algorithm + highly accurate virtual screening sorting algorithm
HPC high-performance computing cluster, supporting screening of molecular libraries at the billion-scale level
Support for four types of molecular generation scenarios
Exploration of chemical space up to 10^19

Hit to Lead

Optimization based on seed molecules with generative AI
Instantaneous support for 50+ ADMET property predictions within seconds.

Potential Target Proteins Exploration

Decoding protein structure and function from sequences.
Probing alterations in protein-protein interactions (PPIs) induced by mutations.

Drug repurposing

Drug repurposing through integration with genomics data.
Probing synergistic dual-drug combination strategies.

AI Capabilities Empowering Life Science Comprehensively

Application

Learn more

Small Molecule Drug Discovery
Hit DiscoveryHit to LeadPotential Target Proteins ExplorationDrug Repurposing
Peptide Drug/Biologics Discovery
Protein Structure PredictionPeptide Drug DesignTherapeutic Antibody Design
mRNA Therapeutics Discovery
Vaccine DesignRNA Secondary Structure Prediction
More
Enzyme EngineeringPlant BreedingGene Therapy

AI-Powered Capabilities

Learn more

Small Molecule
ADMET PredictionDrug Synergy PredictionDrug RepurposingRetrosynthetic Planning
Protein
Protein Structure PredictionProtein Complex Structure PredictionProtein Function Prediction
小分子蛋白
Virtual ScreeningDTA PredictionDrug Design
蛋白蛋白
Ag-Ab Structure PredictionAg-Ab Affinity PredictionMutation-driven PPI Change PredictionPeptide Drug DesignTherapeutic Antibody Design
mRNA
mRNA Sequence DesignRNA Secondary Structure Prediction5’ UTR Design

WENXIN·Foundation Model for Science

Guided by the synergy of "data" + "principles"

Learn more

PaddlePaddle Deep Learning Framework

Flexible Collaboration Modes

AI-powered Comprehensive Solution

Offer an all-in-one solution, empowering pharmaceutical companies, CRO/CDMO, and Biotech with comprehensive intelligent and digital capabilities.

Web-based solution

Offers algorithms and computing power. Accessible upon registration, providing a user-friendly interface and a pay-as-you-go model for maximum user flexibility and convenience.

Private Deployment Solution

Offer dockers and relevant documentation to meet your confidentiality needs. Our solution supports deployment in your proprietary hardware environment.

Pipeline Service

Tailored for the pharmaceutical sector, we offer end-to-end pipeline services where clients pay based on delivered results. No need to purchase any models or machines.

More

Contact us for more details

Case Study

Sanofi

Vaccine Design

French pharmaceutical company Sanofi has entered into an agreement with Baidu to utilize their mRNA sequence design algorithm, LinearDesign. This collaboration aims to optimize the design and development of mRNA vaccines and drugs, expediting the development of vaccine and therapeutic products for human diseases, including COVID-19. This partnership underscores Sanofi's confidence in Baidu's PaddleHelix algorithm and will drive innovation in mRNA-based vaccine and treatment approaches.

AIxplorerBio

Hit to Lead

AlxplorerBio has seamlessly integrated HelixADMET foundation model into its exclusive drug platform. This integration has facilitated the successful identification of three promising PCC molecules, markedly amplifying their research and development efficiency. This accomplishment serves as a compelling testament to the practical application of PaddleHelix AI technology within the realm of biomedicine, suggesting its capacity to offer even more comprehensive algorithmic tools and technical solutions for forward-looking pharmaceutical enterprises and research institutions in the days ahead.

LeadBioTK

Hits Discovery

Recognizing the limitations of current breast cancer treatments, LeadBioTK has collaborated with Baidu PaddleHelix to introduce an innovative strategy for developing novel mechanisms in breast cancer therapeutics. Utilizing HelixVS for virtual screening, they have successfully identified six highly promising molecules. These molecules possess the capacity to impede the growth of breast cancer cells by disrupting the protein-protein interaction (PPI) between CDK4/6 and CCND proteins, ultimately deactivating CDK4/6 kinases. This breakthrough offers a promising avenue for advancing new medication development in the field.

Institute of Urban Agriculture,Chinese Academy of Agricultural Sciences

Protein Structure Prediction

The research team discovered three MADS-box family genes in Phalaenopsis orchids. In order to understand the impact and underlying mechanisms of these genes/proteins on orchid development, the team collaborated with the Chengdu Supercomputing Center and utilized Baidu's protein structure prediction model deployed on the supercomputer for their research. The results revealed the significant roles played by these newly discovered genes in orchid development. The related findings were published in the SCI academic journal 'Frontiers in Plant Science'.

Shanghai Institute for Advanced Study, Zhejiang University

Retrosynthetic Planning

Backed by the Shanghai AI Innovation Center and Baidu's funding, a collaboration was formed with Assistant Professor Ying Wei from the Department of Computer Science at City University of Hong Kong. This collaboration effectively addressed complex challenges related to single-step/multi-step retrosynthesis representation, search, and prediction. Presently, these accomplishments are accessible as a service via PaddleHelix platform. The platform is undergoing continuous updates and refinements. Furthermore, additional research papers pertaining to retrosynthesis will be released in the future, fostering the advancement and flourishing of domestically cultivated AI technology and its application ecosystem.

Partners

HelixFold 3: The World's First Complete Replication of AlphaFold 3, Supported by Baidu AI Cloud CHPC to Power Human Life Science Exploration

Technical Updates

2024/09.12

Baidu's HelixDock Full-Atom Diffusion Model Rivals AlphaFold 3; Code and Data Fully Open-Sourced

Technical Updates

2024/05.24

Baidu's antigen-antibody structure prediction model significantly outperforms AlphaFold3.

Technical Updates

2024/05.20

、、、

Discovery of pyrazolo[1,5-a]pyrimidine derivatives targeting TLR4−TLR4∗ homodimerization via AI-powered next-generation screening

Yao-Yao Jiang, Shuai-Ting Yan, Shan-Zhuo Zhang, Meng Wang, Wei-Ming Diao, Jun Li, Xiao-Min Fang, Hang Yin

2025/01.14

RNAErnie、Pretrained Language Models、Nucleotide Sequence Analysis、Fine-tuning、RNA Types

Multi-purpose RNA language modelling with motif-aware pretraining and type-guided fine-tuning

Ning Wang, Jiang Bian, Yuchen Li, Xuhong Li, Shahid Mumtaz, Linghe Kong & Haoyi Xiong

2024/05.13

Protein Function Prediction

Protein Function Prediction with Primary-Tertiary Hierarchical Learning

Arnold Kazadi, Ming Zhang, Jingbo Zhou

2023/12.05

Protein structure prediction、Protein language model、HelixFold-Single

A method for multiple-sequence-alignment-free protein structure prediction using a protein language model

Xiaomin Fang, Fan Wang, Lihang Liu, Jingzhou He, Dayong Lin, Yingfei Xiang, Kunrui Zhu, Xiaonan Zhang, Hua Wu, Hui Li & Le Song

2023/10.09