[ad_1]
W MIT researchers can use machine learning to find potential drugs faster. After training, technology can better predict which molecules have the desired properties that systems can do now. Scientists report this on the MIT website
The model was formed with a dataset containing data from 250,000 molecules with various properties. The system must use the data to predict how best to optimize the molecules if you want to use them as drugs.
Making Medicines
In development, the laboratory often selects a basic molecule with all the desired properties. For example, the molecule must bind to a certain receptor in the body. This basic molecule is then improved, to ensure that it has important properties for drugs. For example, it must be easy to make and dissolve in liquids.
After training, the system is better able to optimize these molecules than current systems. The system can improve the synthesizability and solubility, but also better find the best base molecule.
Scientists now want to test the system for even more properties, which are also relevant. However, more data is needed, while there is less data. Scientists see it as a challenge to create a system that can work with a limited amount of data.
[ad_2]
Source link
