Measuring whole-canopy photosynthesis under drought conditions (Credit: Peggy Greb, USDA ARS)
GENOMIC PRECISION

Climate Resilience

Engineering crops that withstand drought, salinity, and extreme heat — at exascale.

Image: Peggy Greb, USDA ARS

From Genome to Field, at Petascale

Traits like drought tolerance or heat resistance are rarely governed by a single gene — they emerge from networks of thousands of interacting genes, expressed differently under stress. Finding the handful of genetic combinations that matter, out of billions of possibilities, is a search problem that only high-performance computing can solve at the scale modern breeding programs require.

The HPC Architecture Behind Climate-Resilient Breeding

A simplified view of the computational pipeline that turns raw genomic and environmental data into breeding decisions:

Data Sources Genome sequencers Field & climate sensors Data Ingestion Parallel file systems Petabyte-scale storage Exascale HPC Cluster CPU nodes: GWAS statistics GPU nodes: explainable AI Mixed-precision computing Breeding Targets Candidate genes Ranked crosses Field Trials & Phenotyping Drone & sensor validation of predicted traits Results feed back into the model Continuous feedback loop Each cycle narrows millions of genetic candidates down to a shortlist worth testing in the field.

Simplified data flow: raw genomic and environmental data becomes ranked breeding candidates through exascale computation, validated and refined through field trials.

TOP500 · Rank #3 (June 2026)

Who Works on This: Frontier at Oak Ridge National Laboratory

Frontier, the exascale supercomputer at the Oak Ridge Leadership Computing Facility (ORNL, Tennessee, USA), is currently ranked No. 3 on the TOP500 list of the world's fastest supercomputers, with 1.353 Exaflop/s on the HPL benchmark — behind China's LineShine and the U.S. system El Capitan.

Computational biologist Dan Jacobson and his team at ORNL use Frontier — and previously the Summit supercomputer, on which they won an ACM Gordon Bell Prize — to run explainable-AI genomic selection algorithms across "polytopes" of multi-omics data, searching for the genetic combinations behind climate adaptation in crops such as poplar, switchgrass, and sorghum.

Voices from the Field

Dan Jacobson

Oak Ridge National Laboratory (ORNL)

On designing explainable-AI algorithms for climate-resilient crops on Summit and Frontier.

Read interview
Dan Jacobson

ORNL, via ACM interview series

A deeper technical interview on the explainable-AI genomic selection method and "climatypes."

Read interview
Ana Caño-Delgado

Center for Research in Agricultural Genomics (CRAG), Barcelona

On engineering drought resistance in plants without sacrificing growth, via brassinosteroid signaling.

Read article
Renzo Bonifazi

Wageningen University & Research

On balancing genomic selection speed against loss of genetic diversity — a lesson from livestock breeding relevant to crops too.

Read article