An artificial intelligence from Google unravels a million-year-old mystery in minutes

Una Inteligencia artificial de Google desvela en minutos un misterio de millones de años

For more than half a century Humanity faces the challenge of discovering the shape of the basic building blocks of life, A knowledge To help cure diseases There is no solution today. In 1969 One biologist estimated that it would take more than 14 billion years to discover the possible configurations of a protein. But the system Alphafold was able to do it in a few minutes.

“Instructions” for more than 20,000 types of proteins are stored in human DNA Who is responsible Other functions include mimicking a genetic material, gaining energy, or developing antibodies. Biologists are trying Understand their structures In a process that can take months or years, but this artificial intelligence system owned by Google is capable of doing it in minutes.

This system, is called Predicts the structure of proteins after studying long chains of molecules that make up alpha, As you can predict what a recipe will look like by reading the material. The Deep Mind and the European Molecular Biology Lab, which developed the system, published this Thursday’s issue. ‘Nature’ More than 350,000 protein structures.

Unprecedented progress in biology

Venki Ramakrishnan, a scientist who won the Nobel Prize in Chemistry in 2009, confirms this when he collects ‘Country’, what is that “An exciting breakthrough”, with unpredictable consequences. “It happened earlier than many experts had predicted. It will be exciting to see There are many ways in which biological research can radically change. “ The scientist said in a statement.

These structures can be used to explore treatments for diseases such as Alzheimer’s, cancer and COVID. Because they are all related to the shape of certain proteins. The company’s director Demis Hasabis announced it In the coming months, it will release an estimated 100 million structures with this alphafold system.

Prediction with errors

Spanish biologist Jose Antonio Marquez explained ‘Country’ That Clarify the shape of a protein With a synchronization or alternative method of electron cryomicroscopy, It can take “months or years”, When Despite the bugs this new system does it in minutes. “About Computer predictions, Not from the test resolution of the structure. And the accuracy is 58%, ”explains the biologist.

The scientist promises, though it is not accurate, This system will save a lot of time on investigations. In addition, it also indicates that there is another limitation since LProteins tend to interact with others, and alphafolding predicts isolated molecules, Therefore, it is not capable of predicting cell-cell interactions. “This will accelerate discoveries in all areas of biology. “ Judgment of Marquez.

Leave a Reply

Your email address will not be published. Required fields are marked *