![](//www.4rfv.com/images/no_image.jpg)
A finalist for the Gordon Bell special prize for high performance computing-based COVID-19 research has taught large language models (LLMs) a new lingo - gene sequences - that can unlock insights in genomics, epidemiology and protein engineering.
Published in October, the groundbreaking work is a collaboration by more than two dozen academic and commercial researchers from Argonne National Laboratory, NVIDIA, the University of Chicago and others.
The research team trained an LLM to track genetic mutations and predict variants of concern in SARS-CoV-2, the virus behind COVID-19. While most LLMs applied to biology to date have been trained on datasets of small molecules or proteins, this project is one of the first models trained on raw nucleotide sequences - the smallest units of DNA and RNA.
We hypothesized that moving from protein-level to gene-level data might help us build better models to understand COVID variants, said Arvind Ramanathan, computational biologist at Argonne, who led the project. By training our model to track the entire genome and all the changes that appear in its evolution, we can make better predictions about not just COVID, but any disease with enough genomic data.
The Gordon Bell awards, regarded as the Nobel Prize of high performance computing, will be presented at this week's SC22 conference by the Association for Computing Machinery, which represents around 100,000 computing experts worldwide. Since 2020, the group has awarded a special prize for outstanding research that advances the understanding of COVID with HPC.
Training LLMs on a Four-Letter Language LLMs have long been trained on human languages, which usually comprise a couple dozen letters that can be arranged into tens of thousands of words, and joined together into longer sentences and paragraphs. The language of biology, on the other hand, has only four letters representing nucleotides - A, T, G and C in DNA, or A, U, G and C in RNA - arranged into different sequences as genes.
While fewer letters may seem like a simpler challenge for AI, language models for biology are actually far more complicated. That's because the genome - made up of over 3 billion nucleotides in humans, and about 30,000 nucleotides in coronaviruses - is difficult to break down into distinct, meaningful units.
When it comes to understanding the code of life, a major challenge is that the sequencing information in the genome is quite vast, Ramanathan said. The meaning of a nucleotide sequence can be affected by another sequence that's much further away than the next sentence or paragraph would be in human text. It could reach over the equivalent of chapters in a book.
NVIDIA collaborators on the project designed a hierarchical diffusion method that enabled the LLM to treat long strings of around 1,500 nucleotides as if they were sentences.
Standard language models have trouble generating coherent long sequences and learning the underlying distribution of different variants, said paper co-author Anima Anandkumar, senior director of AI research at NVIDIA and Bren professor in the computing + mathematical sciences department at Caltech. We developed a diffusion model that operates at a higher level of detail that allows us to generate realistic variants and capture better statistics.
Predicting COVID Variants of Concern Using open-source data from the Bacterial and Viral Bioinformatics Resource Center, the team first pretrained its LLM on more than 110 million gene sequences from prokaryotes, which are single-celled organisms like bacteria. It then fine-tuned the model using 1.5 million high-quality genome sequences for the COVID virus.
By pretraining on a broader dataset, the researchers also ensured their model could generalize to other prediction tasks in future projects - making it one of the first whole-genome-scale models with this capability.
Once fine-tuned on COVID data, the LLM was able to distinguish between genome sequences of the virus' variants. It was also able to generate its own nucleotide sequences, predicting potential mutations of the COVID genome that could help scientists anticipate future variants of concern.
Trained on a year's worth of SARS-CoV-2 genome data, the model can infer the distinction between various viral strains. Each dot on the left corresponds to a sequenced SARS-CoV-2 viral strain, color-coded by variant. The figure on the right zooms into one particular strain of the virus, which captures evolutionary couplings across the viral proteins specific to this strain. Image courtesy of Argonne National Laboratory's Bharat Kale, Max Zvyagin and Michael E. Papka. Most researchers have been tracking mutations in the spike protein of the COVID virus, specifically the domain that binds with human cells, Ramanathan said. But there are other proteins in the viral genome that go through frequent mutations and are important to understand.
The model could also integrate with popular protein-structure-prediction models like AlphaFold and OpenFold, the paper stated, helping researchers simulate viral structure and study how genetic mutations impact a virus' ability to infect its host. OpenFold is one of the pretrained language models included in the NVIDIA BioNeMo LLM service for developers applying LLMs to digital biology and chemistry applications.
Supercharging AI Training With GPU-Accelerated Supercomputers The team developed its AI models on supercomputers powered by NVIDIA A100 Tensor Core GPUs - including Argonne's Polaris, the U.S. Department of Energy's Perlmutter, and NVIDIA's in-house Selene system. By scaling up to these powerful systems, they achieved performance of more than 1,500 exaflops in training runs, creating the largest biological language models to date.
We're working with models today that have up to 25 billion
Most recent headlines
17/02/2025
TVBEurope talks to the team who brought a piece of broadcasting history back to life on BBC Wales last night
By Jenny Priestley
Published: February 14, 2025 ...
17/02/2025
Eligible studios will receive the relief directly from their local authorities, with no need to submit a claim
By Matthew Corrigan
Published: February 17, 20...
17/02/2025
As DEI initiatives come under pressure in the USA, supporters of Rise, which promotes diversity in the media and entertainment sector, discuss the continuing im...
17/02/2025
EIB Global loan will partially fund SES's continued O3b mPOWER medium earth orbit (MEO) fleet expansion enabling additional capacity worldwide. O3b mPOWER s...
15/02/2025
At Nielsen, Business Resource Groups (BRGs) play a critical role in nurturing a ...
15/02/2025
Innovation to empower creativity: Explore the latest in Lightroom
Rob Christensen February 15, 2025
0 Comments
Whether you want to remove distracting ...
15/02/2025
Wisycom, a worldwide leader in RF technology and provider of wireless solutions, announces a new distribution agreement with EQUAPHON. An Argentinian profession...
15/02/2025
DPA Microphones, the leading manufacturer of high-quality microphone solutions, is pleased to welcome Nicholas Nick Mariano and Vincent Vince Divine to the ...
15/02/2025
Tuxera, a leading provider of quality-assured file systems and networking solutions, will showcase how Tuxera Fusion SMB enables unprecedented performance for m...
15/02/2025
Bitmovin's latest Video Developer Report indicates an increasing role of AI, the slow adoption of next-generation codecs, and continuing challenges when it ...
15/02/2025
When Michelin Enterprises set out to promote its prestigious Hotel Guide in the new Sleepless in the Kitchen , campaign, veteran German/Slovenian gaffer Robert...
15/02/2025
Formula E, the world's first all-electric racing series, revolutionizing motorsport with cutting-edge technology and a commitment to sustainability has sele...
15/02/2025
Agile Content, a global leader in TV and video technology solutions, announces its rebranding to AgileTV.
The transition to AgileTV stems from a commitment to ...
15/02/2025
Hiltron Communications will promote the latest additions to its range of satellite communication products, systems and supporting services at GovSatCom 2025. Th...
15/02/2025
Following the success of the original Airglow Booklight, DoPchoice debuts a new addition to their inflatable line the Float. This lightweight, inflatable, round...
15/02/2025
LAS VEGAS Marshall Electronics has announced that it will debut its newest camera controller, the RCP Plus, at the 2025 NAB Show at Booth N2649 between April 5 ...
15/02/2025
ATLANTA Gray Media has named Blake Sebo as the next general manager of KTUU and KYES, Gray's NBC and CBS affiliates in Anchorage, Alaska....
15/02/2025
KOKKEDAL, Denmark DPA Microphones will feature its new CORE+ technology for distortion-free microphone sound at the 2025 NAB Show, April 5-9, in Las Vegas....
15/02/2025
SWANSEA, U.K. QuickLink will launch StudioPro-NDI and StudioPro-3, two new models of its QuickLine StudioPro video production platform, at the 2025 NAB Show, Ap...
15/02/2025
Live From 2025 NBA All-Star Game Weekend: After 25 Years, Turner Sports All-In F...
15/02/2025
NASCAR Driver Cam on Max Embraces Cloud-Based Workflow for 2025 Season Chris Brown of WBD Sports says Mobii will play key part in delivering signals from the tr...
14/02/2025
Last month, we officially launched the Spotify Partner Program, a powerful new m...
14/02/2025
From an island off the coast of Lutruwita to the global stage in France SBS & ...
14/02/2025
The future of electro-optical/infrared (EO/IR) land systems is evolving so that ...
14/02/2025
Here are 5 reasons why you should attend ISE 2025
Discover more about Calrec: Learn more about Calrec's legacy of innovation in broadcast audio solutions ...
14/02/2025
World Radio Day: On the same wavelength
By Graham Murray, International Sales Manager at Calrec
To mark World Radio Day, Calrec's International Sales Ma...
14/02/2025
DALLAS and TYSONS, Va. Tegna has concluded a broadcast rights agreement that makes Tegna's KFAA in Dallas the exclusive local television partner of the Dall...
14/02/2025
Amazon's Prime Video has released a new app for Apple TV that the streamer said delivers an improved streaming experiences with faster scrolling, improved s...
14/02/2025
CLEVELAND Telos Alliance announced a software update for its Axia Quasar XR and SR line of audio-over-IP (AoIP) mixing consoles and will show the offering at th...
14/02/2025
Puget Systems Returns to HPA Tech Retreat 2025 with New Analysis Tools for Puge...
14/02/2025
Glass Animals: Tour of Earth Uses Blackmagic Live Production Workflow
Brie Clayton February 14, 2025
0 Comments
80six deploys Blackmagic URSA Broadcas...
14/02/2025
A dispute between Paramount and Youtube TV that would have pulled CBS stations and Paramount's cable channels from the streaming service has been temporaril...
14/02/2025
Colourist Deidre (Dee) McClelland details how she helped shape the distinct look and feel of whimsical Oscar-nominated stop-motion film, Memoir of a Snail
By C...
14/02/2025
BitFire said it intends to introduce forward-looking services that simplify the launch of live production workflows, and enhance real-time collaboration
By Jen...
14/02/2025
The claim by the liquidators is the latest twist in the ongoing saga of the companys collapse at the end of 2021
By Matthew Corrigan
Published: February 14, ...
14/02/2025
A carriage dispute between YouTube TV and Paramount could come to a head today (Feb. 13), with the virtual pay TV provider threatening to pull all of the progra...
14/02/2025
ENGLEWOOD CLIFFS, N.J. LG Electronics USA has opened a new in-house, state-of-the-art virtual production studio at its 360,000-square-foot North American headqu...
14/02/2025
NEW YORK In a bid to help local broadcasters find new ways of monetizing their content, the Coalition for Innovative Media Measurement (CIMM) and TVB have relea...
14/02/2025
WASHINGTON Three Democratic U.S. Senators have sent a stern letter blasting FCC Chair Brendan Carr for the weaponization of the agency for attacking broadcast...
14/02/2025
Years after launching the streaming service Apple TV+ in November of 2019, Apple has finally released a version of its steaming app for Android devices....
14/02/2025
FOOTHILL RANCH, Calif. Red Digital Cinema will make its V-RAPTOR [X] and KOMODO-X camera systems available with Nikon's premier Z Mount, providing filmmaker...
14/02/2025
DAYTONA, Fla. The Fox Sports production of this year's Daytona 500 will see the addition of a few new tech elements and the continued unfolding of the broad...
14/02/2025
Watch Bryson Battles Blind Audition on The Voice Get a Four-Chair Turn Michael Bubl said the Boston Conservatory at Berklee musical theater senior might have...
14/02/2025
2025 Daytona 500: FOX Sports Looks To Ride Super Bowl LIX Momentum Into NASCAR S...
14/02/2025
14 Feb 2025
VEON's Kyivstar and the Come Back Alive Foundation Raise USD 2....
14/02/2025
Max Renews THE PITT, Starring Noah Wyle, For A Second Season
Max
Warner Bros. Television Group
The Pitt
New Episodes Of The Medical Drama From John...
14/02/2025
This year, Technicolor proudly celebrates 110 years of shaping the magic of cinema pioneering color, visual effects, and innovation that bring unforgettable l...
14/02/2025
NBA All-Star: More Games, More Teams, More Music, More Sound From the Court Turner Sports' team will deploy familiar tech to capture all the audio By Dan D...
14/02/2025
LIV Golf Continues To Innovate With Drones, Data-Driven Graphics, and All-Access...