With continuous enhancement of deep learning technology, artificial intelligence (AI) is constantly expanding its boundaries. In the medical field, the application of AI is often manifested in assisting diagnosis of early tumors, or helping complete the electronic medical record. But in addition to that, AI can also play a significant role in drug discovery, which traditionally relied heavily on random discovery in the past. The combination of AI and experimental verification (hybrid-method) can accelerate the development of a new generation of targeted therapy drugs, and can also provide more alternative treatment options for cancer patients.
The purpose of patients using drugs is to change the conformation of disease-causing proteins so that they can be degraded and eliminated. To obtain the conformational information of proteins, the traditional method is through experiments. However, the limitations of this method are obvious. On the one hand, it is difficult to cultivate protein crystals. On the other hand, the experimental results are not a fit for all and are only obtained under certain conditions.
With the joining of AI, things can be different. By using machine learning methods to learn the protein and nucleic acid X-ray crystal diffraction and NMR structure data in the Protein Data Bank (PBD) uploaded by researchers from around the world, the AI-powered drug discovery platform can accurately figure out the exact shape and conformation of the disease-causing proteins.
At present, AI & Medicine’s drug and target protein binding mode prediction (docking) is significantly higher in accuracy and speed than other counterparts in the field of drug design. The reason why its predictions are more accurate lies in the use of data-based algorithms. Compared with conventional methods that are largely based on experience and assumptions, it has a clear advantage when the amount of data is large enough.
Since rituximab was approved by the FDA in 1997, there have been more than 40 targeted therapies on the market. But they only cover a dozen of targets. The number of cancer-causing genes is in thousands, so there is a lack of medicines. Scientists have put forward the idea of “developing new treatment potentials with old medicines”. For example, the failed chemotherapy drug Azidothymidine was also used to treat HIV infection. Commonly used drugs may also act on cancer targets.
To match the drug and the target usually needs to test again and again through experiments. Thanks to the increase in computing power and machine learning, a new possibility was there in solving this problem, that is, to replace the natural scientific process of random discovery with algorithms. Through certain algorithm, the platform can match more than 1,400 FDA-approved drugs with about 10,000 potential targets.
Calculations found that among the FDA-approved drugs, few of the priority targets of drugs are the targets they first designed. If no current medication is available for patients, perhaps results obtained from the algorithm can provide some new options. However, the verification of the corresponding relationship between drugs and targets still requires a lot of experiments. At least, this gives hope to patients who otherwise have no treatment options.