Neo-1
Discover how Neo-1, VantAI's groundbreaking AI model, revolutionizes molecular design and drug development with precise structure predictions and innovative features.
English
English
What is Neo-1 ?
Neo-1 is the world's first AI model launched by VantAI to unify molecular generation and atomic structural prediction. Through innovative technologies, it can predict the structure of biomolecules and generate completely new molecules, especially in the design of new therapeutic drugs such as molecular glues. Neo-1 uses multimodal inputs, which can process partial sequences, partial structures and experimental data at the same time, thereby greatly improving the efficiency and accuracy of drug design.
The core functions of Neo-1
1. Unification of molecular generation and structural prediction: Neo-1 is the first model to combine de novo molecular generation with atomic structural prediction. Unlike traditional models, Neo-1 can not only predict molecular structure, but also create entirely new molecules by generating potential representations of molecules.
2. Multimodal input: This model supports a variety of data inputs, including sequence, structure and experimental data, allowing it to handle more complex tasks and enhances the flexibility and applicability of the model.
3. Large-scale training: Neo-1 is a massive biological diffusion model that uses hundreds of NVIDIA H100 GPUs to train on a wide range of data sets, ensuring its efficiency and accuracy.
4. Customized tools and data sets: Neo-1 combines VantAI's own NeoLink data set and the PINDER & PLINDER tool developed in collaboration with NVIDIA to provide more customized functions and further improve performance.
The technical principles of Neo-1
1. Diffusion process and latent space: Neo-1 transfers the diffusion process from traditional coordinate space to latent space, which allows the model to reason in smoother sequence and structural landscapes, generate new molecules, and predict their structure with atomic precision.
2. Accurate molecular generation: Unlike traditional autoregressive models, Neo-1 adopts a "coarse to fine" generation method, which can accurately guide the generation process of molecules by applying intermediate rewards.
Neo-1 application scenarios
1. Molecular glue design: Neo-1 can accelerate the design of molecular glue for complex targets and significantly shorten the drug development cycle.
2. Protein complex structure prediction: This model can predict the structure of multiple biomolecular complexes and help study the interaction between antibodies and antigens.
3. Combined with the NeoLink data platform: Through integration with VantAI's NeoLink data platform, Neo-1 can assemble a complete atomic-level resolution structure based on structural constraints generated by crosslink mass spectrometry.
4. Antibody discovery: Neo-1 can also achieve rational antibody discovery from beginning to end, which is particularly suitable for the design of new antibody drugs.
Advantages of Neo-1
Neo-1 provides an innovative and efficient way to accelerate drug design and biomolecular research through its advanced AI technology. Its ability to combine generative models and structural prediction makes the discovery and application of biomolecules more accurate. Whether it is drug development, protein complex analysis, or antibody discovery, Neo-1 can provide high-quality predictions and production results in a short period of time.
Visit the official website of Neo-1 project
For more information, please visit Neo-1 's [official website] ( https://www.vant.ai/Neo-1?utm_source=ai-bot.cn ).
