Scientists of the pharmaceutical school Chapel Hill Eshelman based at the University of North Carolina have developed a system called ReLeaSE, which can create molecules of drugs "from scratch". This can speed up the emergence of new medicines.
ReLeaSE (Reinforcement Learning for Structural Evolution) is a computer program consisting of two neural networks, which can be conditionally called a "student" and a "teacher". The algorithm works as follows:
- "Teacher" knows the properties and characteristics of the interaction of more than 1.7 million biologically active molecules and shares this information with the "student".
- "Student" in the process of mastering knowledge offers new molecules that can be used to create medicines.
- "Teacher" approves an effective molecule, laying down information about it in the memory of the "student", preventing similar mistakes in the future.
If we compare this process to learning a language, then after the student learns the molecular alphabet and the rules of the language, they can create new 'words,' or molecules. If the new molecule is realistic and has the desired effect, the teacher approves. If not, the teacher disapproves, forcing the student to avoid bad molecules and create good ones.
The team of scientists has already been able to generate molecules with the desired properties (desired bioactivity, safety profiles) and individual physical characteristics (melting point, water solubility, enzyme effect) using ReLeaSE.