Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics

Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics
Imagen de Editorial Team
porEditorial Team
Argentina

The companies, together with EMBL-EBI, generated open three-dimensional models of viral proteomes to accelerate research in response to future health outbreaks

Add La Derecha Diario on
Share:

Nvidia, Google DeepMind, and the European Bioinformatics Institute of the European Molecular Biology Laboratory generated an open set of protein structures for over 2,800 viruses. The material, available in the AlphaFold database, includes three-dimensional models of 2,812 viral proteomes belonging to 23 families relevant to human health.

The central goal is for laboratories in any country to study potential health threats without waiting for an emergency to be declared. The World Health Organization warns that climate change, international mobility, human expansion into animal habitats, and gaps in surveillance increase the risk of new outbreaks or pandemics.

Proteins allow a virus to enter cells, multiply, and evade the body's defenses. Knowing their three-dimensional shape and how they interact with each other is key to understanding the functioning of a pathogen and designing faster responses.

The predictions were made using AlphaFold2, Google DeepMind's artificial intelligence system that calculates the spatial structure of a protein from its amino acid sequence. Nvidia contributed its BioNeMo platform to process thousands of viral proteins on a large scale.

Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics

Limitations and Available Data

The models do not replace laboratory experiments. The database indicates the confidence level of each prediction, and researchers must validate the results before moving forward with treatments. Nevertheless, the information allows for deciding which proteins to prioritize in the early days of an outbreak, when time is crucial.

The collection includes 5,279 high-confidence heterodimers—complexes formed by two distinct proteins—and 2,749 homodimers, which group two copies of the same protein. It also incorporates another 4,681 high-confidence viral homodimers developed by an external team.

During the Covid-19 pandemic, research on diagnostics, vaccines, and drugs advanced at an unprecedented speed.

The material is hosted on the AlphaFold Pandemic Preparedness Portal, where scientists can consult structures of viruses linked to families that infect humans and download the information for their own analyses. The license permits academic and commercial uses with proper attribution.

Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics

International Collaboration and Global Context

The alliance includes the Coalition for Epidemic Preparedness Innovations, Seoul National University, the University of Glasgow, the Swiss Institute of Bioinformatics, and other institutions. Nvidia also released the tool used to generate the predictions, allowing other teams to analyze new viral sequences.

“Making this data available is essential for understanding viral diagnostics and developing treatments and vaccines,” said Jo McEntyre, interim director of EMBL-EBI, according to Nvidia's blog.

The publication coincides with the high-level meeting of the United Nations General Assembly on pandemic prevention, preparedness, and response, scheduled for September 25 in New York. The meeting will review commitments made after Covid-19 and discuss how to improve international coordination.

Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics

AlphaFold is an artificial intelligence tool developed by Google DeepMind that estimates the three-dimensional structure of proteins from the amino acid sequence. For decades, this task relied on slow and costly laboratory methods, such as X-ray crystallography.

The Contribution of Artificial Intelligence

With artificial intelligence, AlphaFold generates these predictions in minutes or hours and, in some cases, achieves levels of accuracy comparable to those obtained through experimental techniques. This advancement allows for a more agile approach to one of the historical challenges of structural biology.

Nvidia and Google DeepMind use AI to map proteins from 2,800 viruses and anticipate pandemics

The public availability of these models represents a concrete step for the global scientific community to have structured information before a new virus becomes a widespread threat. The combination of open data and large-scale computing capacity aims to shorten response times in future scenarios.

By making both the structures and the prediction tool available, the involved institutions aim to strengthen research capacity in regions with less infrastructure. The focus remains on preparedness and not just on reaction once the problem has arisen.



La Derecha Diario logo
TwitterInstagramYouTubeTikTokFacebook
Derecha Diario TV

Nosotros

  • Quienes Somos
  • Autores
  • Donar

Privacidad

  • Protección de datos
  • Canales
  • Sitemap
  • RSS

Contacto

  • info@derechadiario.com.ar
PUBLICIDAD
La Derecha Diario logoLa Derecha Diario logo
ESX logoInstagram logoYouTube logoTikTok logoFacebook
ARGENTINAUNITED STATESECUADORISRAELMEXICOURUGUAY
DERECHA DIARIO TV
  • ES
    XInstagramYouTubeTikTokFacebook
  • DERECHA DIARIO TV
  • Sections
  • ARGENTINA
  • UNITED STATES
  • ECUADOR
  • ISRAEL
  • MEXICO
  • URUGUAY
  • Countries
  • La Derecha Diario logoLA DERECHA DIARIO
  • La Derecha Diario México logoLA DERECHA DIARIO MÉXICO
  • La Derecha Diario Uruguay logoLA DERECHA DIARIO URUGUAY
  • La Derecha Diario Ecuador logoLA DERECHA DIARIO ECUADOR
  • La Derecha Diario Israel logoLA DERECHA DIARIO ISRAEL
  • La Derecha Diario Estados Unidos logoLA DERECHA DIARIO ESTADOS UNIDOS
  • Topics
  • GUERRA EN IRÁN
  • The Newspaper
  • QUIENES SOMOS
  • AUTORES
  • PUBLICIDAD
  • DONAR

Related news

OpenAI acknowledged that AI agents posted 53 user-submitted images on the internet

OpenAI acknowledged that AI agents posted 53 user-submitted images on the internet

Tesla accelerates its arrival in Argentina and would open its first store in October

Tesla accelerates its arrival in Argentina and would open its first store in October

"Revenge in Prison" tops the list of the 10 most-watched movies on Netflix Argentina

"Revenge in Prison" tops the list of the 10 most-watched movies on Netflix Argentina

Bobby Flores and his family suffered a violent robbery at their home in Saavedra

Bobby Flores and his family suffered a violent robbery at their home in Saavedra

Xuper TV: why watching series and movies for free can ruin your phone or Smart TV

Xuper TV: why watching series and movies for free can ruin your phone or Smart TV

Deputies of Axel Kicillof did not vote on a project by Mayra Mendoza and Tignanelli

Deputies of Axel Kicillof did not vote on a project by Mayra Mendoza and Tignanelli