Synthesis and reverse synthesis are important in organic chemistry. The purpose of reverse synthesis is to find reactants and synthetic routes that can synthesize the target compound. By using deep learning technology to learn chemical reaction-based data, artificial intelligence (AI) can help chemists design synthetic routes and select intermediate compounds, thereby greatly reducing the time required for drug design.

Chemical Drug Development
Why is chemical synthesis so important in the entire biomedicine and pharmaceutical fields? As is known, the biomedicine and pharmaceutical R&D process can be roughly divided into two stages and the first is the preclinical stage, which involves the selection of protein targets and the selection and design of lead compounds (including the optimization and screening of lead compounds). After designing a good molecule, researchers can conduct in vitro and in vivo experiments. Finally, if the experimental results are good, then the project can move forward to the next stage, namely clinical stage. Clinical stage can be further divided into clinical stage I, II, and III.

In fact, it often takes an average of 14 years to develop a drug from target selection to approval for marketing. And the pre-clinical stage will take about 5 to 8 years. This is a very long process and long time means huge investment, which can be summarized into a more famous law called Eroom's law.

Eroom's law means that with the passage of time and the development of technology, the efficiency of new drug research and development is getting lower and lower, and the investment is getting higher and higher. Now it costs an average of more than US$2 billion to develop a drug, which is a very huge investment. At the same time, the research and development of new drugs is accompanied by huge risks, and they are getting higher and higher over time, therefore, the success rate of research and development is very low. Also, the high price and long R&D cycle have caused the rate of return on R&D investment to decrease year by year.

AI comes as a rescue to reduce costs and increase efficiency as it has ushered in explosive development in recent years, with breakthroughs made in data, computing power, and algorithms. MedAI uses the long and short-term memory (LSTM) neural network model to help with chemical synthesis. With AI-based algorithm, there are dozens to hundreds of possibilities, and each move will have a great impact on the subsequent strategies. Advanced algorithms can quickly find strategies with greater wins.

Chemical Retrosynthesis
When researcher are designing the target molecule, they gradually decompose the target molecule into intermediate compounds, and then decompose it into structures. In this way, a synthesis path for the target molecule can be quickly found. This is called a reverse synthesis reaction.

On the whole, the reverse synthesis reaction is to continuously decompose the existing molecules that cannot be synthesized or purchased into a series of precursors, so that each precursor can be purchased or synthesized.

This process also involves a huge search space, resulting in computational complexity. Some researchers have divided the entire reverse chemical reaction into two modules. The first module is single-step reverse reaction prediction. Every compound has dozens of synthesis methods, but these dozens of synthesis methods may not be known. A new model is needed to predict molecular precursors. The second module is a multi-step reverse reaction search, and the computer test chemical experiment has achieved very good results.

For moderately difficult molecules, more experienced chemists can make a relatively good reaction path in a few hours to a day, but AI can accelerate the entire process to a few seconds. In addition to chemical drug development and reverse reaction synthesis, AI can be applied in nearly the entire drug development such as target screening, drug design, drug molecule generation, and drug screening.

Author's Bio: