Recently, DeepMind, an artificial intelligence company under Google, opened sourced its Nobel Prize winning drug development AI model AlphaFold 3. This means that from now on, researchers from various fields of biomedicine can download the model code for free and run the model themselves to carry out drug development work.
Public information shows that the Google AlphaFold model can be used to predict protein structure and interactions. The latest version of the model, AlphaFold 3, not only has the ability to predict proteins, but also has the ability to predict the structures and interactions of almost all biomolecules such as DNA, RNA, and small molecules, directly promoting drug discovery and disease treatment.
DeepMind CEO Demis Hasabis commented that AlphaFold 3 is an important milestone. Biology is a dynamic system, and you must understand how physiological characteristics are generated through the interactions between different molecules in cells. You can see AlphaFold 3 as a big step in this direction
Generally speaking, developing drugs through conventional experimental methods takes several years and incurs extremely high costs, but AlphaFold 3 can significantly accelerate this process. It can quickly screen potential drug targets, predict the structure of target proteins, and discover and search for their medicinal pockets.
It is worth mentioning that when AlphaFold 3 was released in May this year, it was widely controversial due to the lack of open underlying code. Scientists believe that this move undermines the reproducibility of research and development. So DeepMind immediately changed its policy and promised to release an open-source version of this AI model within six months. At present, obtaining this model requires filling out a form for application, and once approved by DeepMind, the application rights can be obtained.
AI pharmaceuticals are poised to take off
It cannot be denied that artificial intelligence systems represented by AlphaFold have made significant breakthroughs in the field of life sciences, promoting the application of AI in drug development. At the same time, IT giants see biotechnology as the next frontier for AI applications: last year Salesforce launched the protein generation model ProGen, and earlier Microsoft also released an open-source model EvoDiff similar to AlphaFold.
Focusing on the domestic market, the layout around AI drug development models is also hot. For example, in May this year, the Baidu PaddlePaddle PaddleHelix team developed the HelixFold Multimedia model, which is said to be the industry leader in predicting the structure of antigen antibody/polypeptide protein complex; In recent years, Huawei has also constantly updated Pangu's drug molecular model and combined with the Chinese Academy of Sciences Shanghai Institute of Materia Medica to enable the full process of AI drug design.
According to a research report by Huachuang Securities, AI has shown great potential in the drug discovery stage. AI technology, especially machine learning and deep learning algorithms, can quickly analyze and process large amounts of biomedical data, identify potential candidate compounds, predict their effects and side effects, thereby significantly reducing the time required for drug discovery. According to Nvidia, the use of AI technology can shorten the time required for drug discovery to one-third of the original, and save costs to 1/200.
According to Precedence Research, the Al pharmaceutical industry is expected to maintain rapid growth over the next decade, with a market size of $1.17 billion by 2023 and projected to exceed $11.8 billion by 2032. The CAGR from 2023 to 2032 is projected to reach 29.3%.