When it comes to the research and development of drugs
谈到药物的研发，不可避免的就要谈到蛋白质，因为大部分的药物进入人体之后，都是需要与体内的蛋白质相互作用才能发挥疗效。而蛋白质是一个极其复杂的大分子化合物，虽然只需要20种氨基酸排列组合就能在人体内合成蛋白质，但是蛋白质三维结构的预测却成为了困扰科学界半个世纪的难题，并由此产生了蛋白质结构学这一学科，专门研究和预测蛋白质的结构。
, it is inevitable to talk about proteins, because after most drugs enter the human body, they need to interact with proteins in the body to exert their curative effect. Protein is an extremely complex macromolecular compound. Although only 20 kinds of amino acid arrangements are required to synthesize proteins in the human body, the prediction of the three-dimensional structure of proteins has become a problem that has plagued the scientific community for half a century, resulting in the discipline of protein structure, which specializes in the study and prediction of protein structures.
Until July 15, 2021, Google’s DeepMind team published a paper called “Highly accurate protein structure prediction with AlphaFold” in “Nature”
, borrowing AI technology to achieve the accuracy of experimental determination for the first time, in a sense The problem of protein three-dimensional structure prediction is solved. At the same time, after the publication of this article, the DeepMind team also open sourced the code of AlphaFold, so that this achievement, known as “one of the most important scientific breakthroughs made by mankind in the 21st century”, can be shared by all mankind.
直到2021年7月15日，谷歌的DeepMind团队在《Nature》上发表了名为“Highly accurate protein structure prediction with AlphaFold”的论文，借用AI技术第一次达到了实验测定的精度，在某种意义上将蛋白质三维结构预测的难题所解决。同时在这篇文章发表后，DeepMind团队也开源了AlphaFold的代码，使得这个被誉为“人类在21世纪取得的最重要的科学突破之一”的成就，能够被全人类所共享。
With the great success of AlphaFold in the field of protein structure prediction
, people have begun to pay attention to some of its shortcomings and room for improvement, especially in the direction of drug research and development. Although AlphaFold has brought breakthroughs from zero to one, it is far from assisting drug development. , there is still a long way to go.
随着AlphaFold在蛋白质结构预测领域的巨大成功，人们也开始关注到一些它的一些不足以及进步空间，尤其是在药物研发方向，AlphaFold虽然带来了从零到一的突破，但是距离辅助药物开发，依然有着很长路要走。
First, the result obtained by AlphaFold is still a prediction result
, which is the optimal solution considered by a probability-based AI. Even those results with high confidence still need to be verified by researchers at the experimental level. In the process of drug research and development, it is too early to rely solely on AI predictions for some protein structures that are closely related to human health.
其一，AlphaFold所得到的结果依旧是一种预测结果，是一种基于概率的AI所认为的最优解，即使那些置信度很高的结果，依然需要研究人员在实验层面上加以验证。在药物研发过程中，与人体健康息息相关的一些蛋白质结构，完全依靠AI的预测还言之过早。
Second, AlphaFold’s prediction of protein structure is still in the static stage.
However, in the human body, including in the process of combining with drugs, the protein structure is by no means static, and the dynamic protein structure is sometimes more than static. The structure is more valuable and meaningful, but at the same time, it is more difficult to study.
其二，AlphaFold对蛋白质结构的预测还停留在静态阶段，然而在人体内，包括在与药物相结合的过程中，蛋白质的结构绝不是一成不变的，而动态的蛋白质结构，在某些时候比静止的结构更加具有研究价值和意义，当然与此同时，研究难度也更大。
Third, AlphaFold’s prediction results are relatively unsatisfactory for the structures of those proteins that do not have enough homologous sequences
, which has also brought criticism from some protein structure experts. In drug development, researchers are not limited to those proteins with homologous evolution information, so the predictive ability of AlphaFold in this aspect needs to be improved.
其三，AlphaFold对于没有足够同源序列的那些蛋白质的结构，预测结果相对来说不是很理想，这也带来了一些蛋白质结构学专家对它的诟病。在药物研发中，科研人员并不会局限于那些有同源进化信息的蛋白质进行研究，所以，AlphaFold在这一方面的预测能力，还有待提高。
To sum up, this breakthrough of AlphaFold is not the end of protein structure.
On the contrary, it just provides a tool to help researchers focus more on drug development itself. It opens up AI as a tool. A new era in complementary drug development. At the same time, the emergence of AlphaFold has also created more possibilities for drug research and development, such as the prediction of binding sites between drugs and proteins, such as the improvement of existing drugs, such as the research on the binding degree of drugs and proteins under different environmental conditions, etc. Wait.
综上所述，AlphaFold的这次突破，并不是蛋白质结构学的终结，相反，它只是提供了一种工具，帮助研究人员更加聚焦在药物研发本身，它以一种工具的角色，开启了AI辅助药物研发的新纪元。同时，AlphaFold的出现，也给药物研发创造了更多的可能，比如药物与蛋白质结合位点的预测，比如对已有药物的改进工作，比如在不同环境条件下药物与蛋白质结合度的研究等等。
Regarding AlphaFold and drug research and development
, even if the chemical reaction generated by the collision between the two seems to be unclear, one thing is certain, that is, the concept of AI + medicine has begun to heat up. While AI technology is gradually being applied to the pharmaceutical industry, Some technology companies and capital markets have also begun to pay attention to this field.
关于AlphaFold与药物研发，即使两者碰撞之下产生的化学反应似乎还不是很明朗，但有一点可以确定，那就是AI+医药的概念已经开始升温，在AI技术逐步被应用于医药行业的同时，一些科技公司和资本市场也开始了对这一领域的关注。
