Article Synopsis

  • Researchers have created ESM3, a powerful language model that can generate functional proteins based on over 3 billion years of evolutionary biology.
  • ESM3 processes various aspects of proteins—sequence, structure, and function—allowing it to respond to complex requests and enhance its output accuracy.
  • In tests, ESM3 successfully generated a bright fluorescent protein with only 58% similarity to existing ones, simulating around 500 million years of evolutionary divergence.

Category Ranking

98%

Total Visits

921

Avg Visit Duration

2 minutes

Citations

20

Article Abstract

More than 3 billion years of evolution have produced an image of biology encoded into the space of natural proteins. Here, we show that language models trained at scale on evolutionary data can generate functional proteins that are far away from known proteins. We present ESM3, a frontier multimodal generative language model that reasons over the sequence, structure, and function of proteins. ESM3 can follow complex prompts combining its modalities and is highly responsive to alignment to improve its fidelity. We have prompted ESM3 to generate fluorescent proteins. Among the generations that we synthesized, we found a bright fluorescent protein at a far distance (58% sequence identity) from known fluorescent proteins, which we estimate is equivalent to simulating 500 million years of evolution.

Download full-text PDF

Source
http://dx.doi.org/10.1126/science.ads0018DOI Listing

Publication Analysis

Top Keywords

years evolution
12
simulating 500
8
500 years
8
language model
8
proteins esm3
8
fluorescent proteins
8
proteins
6
evolution language
4
model billion
4
billion years
4

Similar Publications

Objective: Estimate mortality indicators and impact of COVID-19 on healthcare workers in Bahia in the period 2020-2022.

Methods: This is a descriptive study, with death data extracted from the Brazilian Mortality Information System. Population data were obtained from professional councils, the National Registry of Health Establishments and the Brazilian National Immunization Program Information System.

View Article and Find Full Text PDF

Factors associated with excessive screen time in the Brazilian population: a panel study with 254.600 adults and elderly.

Cien Saude Colet

August 2025

Faculdade de Medicina, Universidade Federal do Rio Grande Faculdade de Medicina. R. General Osório s/n, Centro. 96200-400 Rio Grande RS Brasil.

Screen time has prompted investigations by researchers worldwide because of its impact on general health. This research aimed to analyze excessive screen time from a Brazilian national survey among adults and older people and to verify the immediate effect of the COVID-19 pandemic on the evolution of the behavior. A panel study using the survey database between 2016-2022, in a sample of 254,600 Brazilian adults and elderly residents in capital cities.

View Article and Find Full Text PDF

Research over the last 20 years has shed important light on the vocal behaviour of our closest living relatives, bonobos and chimpanzees, but mostly relies on qualitative vocal repertoires, for which quantitative validations are absent. Such data are critical for a holistic understanding of a species` communication system and unpacking how these systems compare more broadly with other primate and non-primate species. Here we make key progress by providing the first quantitative validation of a Pan vocal repertoire, specifically for wild bonobos.

View Article and Find Full Text PDF

Deltaviruses are subviral agents of animals, which, in humans, require a hepadnavirus helper for transmission. The absence of deltavirus-like endogenous viral elements (δEVEs) has prevented an understanding of their evolution in deep time. By screening the representative genomes of all metazoans for endogenous delta antigen-like sequences, we report the discovery of 13 δEVEs in the genomes of five species of termites.

View Article and Find Full Text PDF

The size and composition of local species pools are, in part, determined by past dispersal events. Predicting how communities respond to future disturbances, such as fluctuating environmental conditions, requires knowledge of such histories. We assessed the influence of a historical dispersal event on community assembly by simulating various scales of dispersal for 240 serpentine annual plant communities that experienced a large shift from drought to high rainfall conditions over three years.

View Article and Find Full Text PDF