Xu Jinbo, founder of AI Protein folding: NewOrigin big model will "customize" protein drugs with one click in the future

Tech 2023-07-07 18:11:58 Source: Network

On July 7th, at the 2023 World Artificial Intelligence Conference (WAIC), the industry's first AI protein generation model, "NewOrigin" (also known as "Darwin" in Chinese), was officially unveiled. Professor Xu Jinbo, the research and development leader of NewOrigin big model and the outstanding visiting professor of Tsinghua University Institute of Intelligent Industry and the founder of Molecular Heart, said that the AI protein generation big model aims at the real industrial application needs such as innovative drug design and Synthetic biology, and will use a model to meet the needs of the whole process of protein generation

On July 7th, at the 2023 World Artificial Intelligence Conference (WAIC), the industry's first AI protein generation model, "NewOrigin" (also known as "Darwin" in Chinese), was officially unveiled. Professor Xu Jinbo, the research and development leader of NewOrigin big model and the outstanding visiting professor of Tsinghua University Institute of Intelligent Industry and the founder of Molecular Heart, said that the AI protein generation big model aims at the real industrial application needs such as innovative drug design and Synthetic biology, and will use a model to meet the needs of the whole process of protein generation. In the future, protein design such as macromolecular drugs and new biomaterials can achieve "one click customization".

It is reported that NewOrigin big model can realize multi-mode directional generation by learning hundreds of billions of levels of multi-mode Big data. A single model can meet the needs of the whole process of protein generation, such as sequence generation, structure prediction, function prediction, ab initio design, etc., solve the problem of specific functional protein generation required by industrial applications, and evaluate the effect and value in the real industrial environment.

In the past 60 years, Protein structure has always been one of the ultimate problems puzzling biologists. Scientists have been awarded Nobel Prize for resolving the three-dimensional structure of important proteins such as hemoglobin for many times. This situation has undergone fundamental changes after the application of AI methods. In 2016, Professor Xu Jinbo used AI for the first time to significantly improve the accuracy of Protein structure prediction. Since then, AI has completely subverted protein prediction and gradually affected protein production. However, due to extremely high technological barriers, protein generation remains a technical challenge that global scientists have not yet fully overcome.

"The emergence of large models will greatly accelerate the development process of protein production technology, and promote its application in biomedicine, Synthetic biology and other fields, thus changing the pattern of Bioeconomy." Xu Jinbo said in his speech. The current performance of natural language big models such as ChatGPT has doubled the confidence of various sectors in the mechanism of big models. However, in professional vertical fields such as protein generation, the ability of universal natural language large models is very limited. The reason for this is that the complex data, professional knowledge, and application scenarios in the field of biology are far from the common scenarios of natural language interaction, and the ability requirements are also higher.

Therefore, in order to develop a large model for protein generation, in addition to the necessary basic conditions such as algorithms, computing power, and data, two major professional advanced abilities are also required: one is to integrate multiple disciplines such as computer science, biology, and physics, be familiar with multiple methods such as AI, molecular dynamics, and quantum computing, and be able to consider the cross domain fusion ability of sequence and structure, main and side chains, evolution and omics in parallel in practice; The second is to step out of the laboratory and sink into the real industrial environment, with the ability to meet the real industrial needs in terms of demand, verification, and implementation. A team with these abilities and conditions is very scarce, "Xu Jinbo believes.

Xu Jinbo's team has been developing protein design algorithms using pre training mechanisms since 2019. By integrating various technologies such as structural prediction, side chain prediction, and protein protein docking, combined with various scenario requirements, significant breakthroughs have been made in modifying or redesigning proteins. For example, designing proteins with similar but smaller functions, proteins that can bind to a small molecule, enzymes that can bind to a substrate, proteins for gene editing, etc.

On this basis, Molecular Heart has developed a large-scale AI protein generation model, NewOrigin, which integrates natural language and protein language. It has five advantages: it can customize the generation of proteins according to specific needs, such as generating antibodies against a certain target or generating specific enzymes against a certain substrate, achieving "on-demand customization"; Based on hundreds of billions of multimodal data, NewOrigin can perform multimodal input and output, such as generating protein sequences with specific functions or generating three-dimensional structural information of a certain protein, to meet the needs of different levels of applications; At the same time, in order to evaluate the generation effect, NewOrigin integrates multiple methods such as AI, molecular dynamics, and quantum computing to form a multidimensional feedback mechanism to achieve rapid verification and iteration. In order to reduce the interaction threshold, NewOrigin uses a protein generation mode that integrates natural language interaction to meet the application needs of biologists without AI technology background. More importantly, the NewOrigin big model is an AI protein big model that truly faces industrial needs, which can continuously iterate based on industrial level application feedback to solve real industrial needs.

In order to better meet the needs of applications, Molecular Heart will integrate NewOrigin big model capabilities in the one-stop protein prediction, optimization and design platform "MoleculeOS", and build industrial solutions based on NewOrigin for Drug design, Synthetic biology and other application scenarios, further integrating NewOrigin big model capabilities in Drug design, biological breeding, environmental protection High performance materials and other fields have been widely implemented, driving innovation in multiple fields. For example, through dialogue and interaction, NewOrigin can generate antibodies targeting a specific target, or generate enzymes with substrate specificity.

Xu Jinbo stated that AI and biotechnology are important strategic areas in current global technology competition, and protein technology, as the underlying technology of biotechnology, is the only way to integrate and innovate with AI. The development of AI protein generation model is just the starting point. What is more valuable is to really apply it to the industry, realize programmable and predictable innovative drug design and biological product development, and drive Bioeconomy reform through bottom technology breakthroughs.

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])

Mobile advertising space rental

Tag: Xu Jinbo founder of AI Protein folding NewOrigin big