This article was automatically translated from HIBAPRESS, the Arabic version:
Hibapress
French researchers have developed an innovative method based on hybrid artificial intelligence (AI), making it possible to design tailor -made proteins, opening the way for new health and environmental applications.
Developed by a team of mathematics researchers from the French Research Institute for Agriculture, Food and the Environment (INRAE), this AI combines “learning and reasoning”, said a press release from the Institute.
“This hybrid AI is thus capable of conceiving proteins according to rules learned by deep learning, but also from physics or explained by designers”, specifies the same source.
“Proteins play a central role in the development and functioning of living beings,” said the French Institute, noting that their three -dimensional structure, resulting from the folding of their amino acid chain, determines their biological function.
Understanding and predicting this replication “constitutes an important research issue, in particular to better study certain diseases caused by a bad folding of proteins, including neurodegenerative diseases such as Alzheimer’s and Parkinson,” said it.
Currently, protein design by AI is progressing at “phenomenal speed”. Recently highlighted by the Nobel Prize for Chemistry 2024, this area now allows the design of “completely new proteins, opening up revolutionary possibilities in the field of health and the environment”.
In this context, “INRAE, through sustainable collaboration between IA researchers and molecular modeler, has contributed to these advances through design methods based on a hybrid AI, mixing learning and reasoning, already implemented to design different new proteins, functional, and experimentally characterized”, indicates the Institute.
“Deep learning” is used to extract the rules that govern the design of proteins, by exploiting sequences and natural protein structures, accumulated for decades by biophysicists in protein data bank, “he explains.
“Automatic reasoning”, continues the Institute, is used for “its ability to combine these rules learned with fundamental laws of physics or guidelines of the designer, to identify, very quickly, the different proteins that meet these requirements among the exponential universe of possible proteins”.
