googleDeepMindofAIThe tool network is becoming increasingly dense, not only in predicting biomolecules such as proteins but also delving into the field of protein design. Recently,DeepMindLaunched its first product dedicated to designing novel high-strength protein conjugatesAIsystemAlphaProteoGenerate new protein conjugates for multiple target proteins. Throughout the entire drug development process,AlphaProteoandDeepmindPreviously launchedAlphaFoldWhat positions are they in, and what kind of connections do they have? Compared to previous technologies,AlphaProteoWhat are the highlights of the protein design platform? What is its application prospect? What impact will it have on China's scientific research community and enterprises, and how should we respond?
Frost & SullivanFrost & Sullivan,Zhou Mingzi, Executive Director of Greater China at Frost & Sullivan (hereinafter referred to as 'Frost & Sullivan'), answered21Interviewed by Century Economic Report, for joint discussionAlphaProteoAs the new generationAIThe importance of adjuvant drug design tools.
21Century Economic Report
*Click at the end of the article Read the original text View the full report
Q: According to the introduction,AlphaProteoIt is the first designed specifically for the synthesis of novel high-strength protein conjugatesAISystem. What does this mean for drug research and development?
AlphaProteoThe importance of drug research and development is mainly reflected in shortening the time required for drug development and improving the effectiveness of research and development.AlphaProteoAs the first to be developed specifically for designing novel high-intensity protein conjugatesAIThe system is capable of generating new protein conjugates for a variety of target proteins, especially in the treatment of cancer and diabetic complicationsVEGF-AHistorical progress has been made in the design of protein conjugates. In targeting7In testing for different target proteins,AlphaProteoOutstanding performance was demonstrated, with the designed protein conjugates exhibiting an affinity that is higher than the best available methods3reach300This breakthrough not only greatly improves research efficiency, reducing the laboratory work that used to take years to just a few weeks or even days, but also opens up new possibilities for future drug development.
In addition,AlphaProteoIt also provides greater possibilities for solving more profound biological problems and for drug design. The design principle is to closely attach to viruses or cancer cells, preventing signal transmission between them, thereby disrupting their function and ultimately leading to the death of pathogens or cancer cells. This mechanism offers new strategies for preventing viral infections and treating cancer.
Q: Throughout the entire drug research and development process,AlphaProteoandDeepmindPreviously launchedAlphaFoldWhat positions are they in, and what connections do they have?
AlphaFoldIt is a protein structure prediction tool that has assisted humanity in understanding life sciences and has accelerated drug research and development efficiency, but it cannot create new proteins.AlphaProteoIt can be directly used in the design of new high-strength protein adhesives, accelerating the discovery of protein targets and the process of drug research and development.AlphaFoldAt present, we have accumulated over1Protein prediction structures in billions,AlphaProteoOn this basis, protein databases (PDBA large amount of protein data in was used for training, thereby mastering various molecular binding modes. By providing the target molecular structure and a set of preferred binding positions on the molecule,AlphaProteoCorresponding candidate proteins can be generated.AlphaFoldandAlphaProteoBoth are at the forefront of target discovery and drug research and development, butAlphaProteoFurthermore, its design capability for protein adhesives has advanced at a rapid paceAIThe application progress in target discovery and drug research and development.
Q: Compared with previous technologies,AlphaProteoWhat are the highlights of the protein design platform? What is its application prospect?
AlphaProteoThe highlights of the protein design platform are mainly reflected in the following aspects:1The success rate of protein design has broken records: in the tests7on the target protein,AlphaProteoThe experimental success rate is higher, in wet laboratory tests,9%reach88%The candidate molecule was successfully bound, which is more efficient than other methods5reach100times. Moreover, it is higher than the binding affinity of the current best method3reach300times;2Ready-to-use binder:AlphaProteoIt can generate 'ready-to-use' binders suitable for multiple applications, requiring only one round of medium-throughput screening without further optimization;3Multi-target design capability:AlphaProteoNot only can protein conjugates targeting a single target be designed, but also those targeting multiple targets, including those related to cancer, inflammation, autoimmune diseases, and viruses.
AlphaProteoAs a new breakthrough in the field of proteomics, starting from drug research and development, it can develop into multiple fields such as agriculture and food science, with broad application prospects. Specifically,AlphaProteoIt can accelerate the understanding of biological processes, help discover new drugs, and develop biosensors.AlphaProteoNew solutions may also be developed in controlling cell signal transduction, imaging proteins, cells and tissues, and endowing various effector systems with target specificity. In addition, throughAlphaProteoThe generated protein conjugates can be applied in multiple fields such as drug development, disease diagnosis, cell and tissue imaging, and enhancing crop resistance.
Q: The research results show thatAlphaProteoWhy are these proteins chosen for experiments that can effectively combine multiple proteins related to infections, cancers, inflammations, and autoimmune diseases? Is the choice more practical or is it more likely to leverage the capabilities of this technology?
AlphaProteoThe selection of proteins related to cancer, inflammation, autoimmune diseases, and viral infections as experimental targets is based on the extensive research needs in these fields and their profound impact on disease treatment. In terms of selection, both aspects are considered, but practicality is perhaps the more important factor. From a practical perspective, areas related to infections, cancer, inflammation, and autoimmune diseases require precise drug development and faster progress in drug research and development.AlphaProteoThe application can accelerate the R&D process in this field. In terms of capability utilization, there is currently a relatively in-depth research foundation in the field of proteins related to infections, cancers, inflammations, and autoimmune diseases. Therefore, choosing this area can better verifyAlphaProteoThe technical precision and capabilities. In addition, from practical applications,AlphaProteoIn the experiment, binding agents were successfully designed for a variety of disease-related proteins, including those associated with cancer and diabetic complicationsVEGF-Atarget protein, andBHRF1andSARS-CoV-2spike protein receptor binding domainSC2RBDWait. These success stories prove thatAlphaProteoThe potential in designing high-strength protein conjugates. This also indicatesAlphaProteoIt can greatly reduce the time required for initial experiments involving a wide range of protein binders, accelerating the process of drug design and disease understanding.
Q: How long will it take for this technology to be truly applied on a large scale for drug research and development? What barriers need to be overcome?
Judging from the current research progress,AlphaProteoIt has demonstrated high success rates and affinity for multiple protein targets, but there are still some challenges to large-scale use in drug research and development, at least5 - 10The time span of [a certain number of] years has been achieved to truly realize large-scale application in drug research and development. Firstly, although the system performs excellently in medium-throughput screening, the number of target proteins verified experimentally is relatively small, which implies that there may still be some unpredictable challenges on a broader range of protein targets. Secondly, existing designs rely on the crystal structure of target proteins, and many potential drug targets may not have known high-resolution structures, which poses a challenge forAlphaProteoThe wide application has raised new demands. In addition, althoughAlphaProteoThe generated conjugates exhibit excellent thermal stability, but their long-term stability and effectiveness in complex in vivo environments still require further validation.
In pursuit of technological innovation to driveAlphaProteoIn the application of the system, the research team faces three key challenges: First, it is necessary to deeply optimize the algorithm to extend its applicability to protein targets lacking high-resolution crystal structures, achieving a broader coverage of targets; second, although high-affinity conjugates have been obtained through medium-throughput screening, to ensure the effectiveness and specificity of conjugates targeting complex targets, multiple rounds of screening and optimization are still required; finally, given that existing research is mainly limited to in vitro experiments, ensuring that these conjugates remain stable and effective in an in vivo environment becomes an important issue that must be overcome in the next phase.
Q: The conclusion states, “We believeAlphaProteoIt will provide new solutions for many biological applications, such as cell signaling, imaging proteins, cells and tissues, and target-specificity enhancement. Are these principles the same as those of this research? What are the challenges in the advancement process?
The applications mentioned in the conclusion, such as cell signaling and imaging proteins, are related toAlphaProteoThe principles of protein binding demonstrated in the study are consistent.AlphaProteoThe core principle of designing highly affinity protein conjugates is to use deep learning models to predict and generate binding agents that can bind specifically to targets based on their structure. This binding mechanism is equally important in applications such as cell signaling and imaging, as these processes require highly specific protein interactions.
PromoteAlphaProteoDuring the application of the system, several technical and practical operational challenges need to be overcome: Firstly, for proteins that dynamically change during cell signal transduction,AlphaProteoFurther development is needed to adapt to its variable conformation; secondly, designed imaging proteins must not only possess high specificity but also maintain stability in complex in vivo environments to prevent rapid degradation; in addition, althoughAlphaProteoHigh target affinity has been demonstrated, but to endow more complex targets with specificity and make them useful in different applications, further experimental verification is still required. Therefore, although the principles of various application fields are similar, each field faces unique technical challenges, which necessitate more detailed experiments, optimization, and stricter validation processes.
Q: What impacts will it have on China's scientific research community and enterprises, and how should they respond?
AlphaProteoThe emergence of technology will bring about significant innovation in drug research and development ideas for China's scientific research community and biopharmaceutical enterprises. Firstly, it will greatly accelerate the process of protein design, reduce the cost and time of experimental screening, improve the efficiency of biopharmaceutical research and development, enable enterprises to advance faster into preclinical testing stages, and significantly shorten the time to market for new drugs; moreover, in the design of complex protein conjugates and antibody drugs, computer simulation can partially replace experimental work, offering further opportunities for optimizing research and development costs; looking ahead,AlphaProteoThe application is not limited to drug development but also covers protein engineering in fields such as environmental protection, providing possibilities for enterprises to open up new markets.
In the face of this trend, China's scientific research community needs to strengthen the learning and application of this technology, promote interdisciplinary cooperation, especially in the integration of artificial intelligence and biomedicine. At the same time, enterprises should increase investment in cutting-edge biotechnology, build strong R&D capabilities and international cooperation networks, in order to seize future market opportunities.
Q: What is the current status of open-source development for this technology? Should it be made open-source, and why?
Currently,AlphaProteoSome details and research findings have not been fully open-sourced, especially those related to biosecurity and commercial use. Although certain methods and results have been made public in academic literature, core machine learning models and algorithms have not yet been made available to the public. This limited open-source strategy is mainly due to concerns about technology abuse and security risks. Open-source can promote global scientific research communities to jointly advance technological progress and increase its application breadth; however, due to biosecurity considerations, some high-risk applications may need to be treated with caution to prevent technology abuse.
Some open-source projects may become mainstream, providing some content for universities and other research institutions to use while retaining core technologies. This is a strategy that balances science popularization with commercial interests. This approach can not only promote academic research but also provide financial support for continuous technological innovation. At the same time, by controlling core technologies, it ensures the quality and safety of technology, achieving a win-win situation.
*This interview has been published in 21Century Economic Report Reporter: Yan Shuo, original title: 'Focus on Digital Health | GoogleDeepMindlaunchAlphaProteoProtein design “hitched up”AIExpress Train
