Google DeepMind's AlphaGenome 2 Predicts How DNA Changes Affect Individual Cell Types

Exercise 1
Vocabulary
Repeat each word, its meaning and the example sentence out loud.
unravel
verb/ʌnˈræv.əl/to understand or explain something complex by examining it carefully, piece by piece
The team spent months trying to unravel how the mutation affects protein function across different tissues.
drill down
phrasal verb/drɪl ˈdaʊn/to examine something in progressively greater detail, going deeper into the layers of information
When we drill down into the data, we can see how the variant affects gene expression in cardiac cells specifically.
granularity
noun/ɡrænˈjʊl.ɑr.ə.ti/the level of detail or precision in analyzing or understanding something; the degree to which something is broken into small, specific components
The computational granularity of the model allows us to predict effects at the single-cell level rather than across entire tissues.
computational cost
noun phrase/ˌkɑm.pjəˈteɪ.ʃən.əl ˈkɔst/the amount of computing power, time, or resources required to perform a calculation or run a model
While AlphaGenome 2 offers superior predictions, its computational cost makes it impractical for analyzing every possible genome variant.
feed into
phrasal verb/ˈfid ˌɪn.tu/to contribute to or supply information that supports a larger process, system, or outcome
Researchers hope that AlphaGenome 2's insights will eventually feed into the development of virtual cell simulations for disease modeling.
Exercise 2
Article
Read the article aloud on your own, or listen to it first.
Google DeepMind's AlphaGenome 2 Predicts How DNA Changes Affect Individual Cell Types
Tap any word to see what it means.
Google DeepMind has unveiled AlphaGenome 2, a successor to its genome-predicting AI model released last year, designed to illuminate how genetic variants affect individual cell types with unprecedented specificity.
While every cell in the human body carries the same DNA, different cell types switch different genes on and off in response to their biological role and state. A genetic variant that appears benign in one cell type may cause pathology in another, making this cellular context crucial for understanding disease mechanisms.
AlphaGenome 2 was trained on more than 100 million individual cell profiles across 68 tissues, enabling it to capture these nuanced differences. The model essentially transforms massive cell atlases—vast repositories that catalog individual cells in detailed maps—into a predictive engine for human genetics.
According to DeepMind, AlphaGenome 2 identifies nearly twice as many variants that alter gene expression compared to the previous state-of-the-art approach. This represents a substantial leap in detecting disease-relevant effects that might otherwise escape notice.
Remarkably, the model can make predictions about new cell types and biological states without requiring retraining. It can also identify gaps in experimental data, helping researchers prioritize which directions to investigate further in the laboratory.
For researchers at DeepMind, the breakthrough lies in tying human genetics directly to cell biology. This capability allows scientists to drill down from genetic variants identified in large population studies to their effects at the cellular level, potentially revealing how they connect to disease.
The specificity proves crucial for drug development, since therapeutics target specific proteins in specific cell types. Without this granularity, researchers lack the precision needed to bridge the gap between population genetics and treatment design.
AlphaGenome 2's more detailed predictions come at a computational cost that makes them considerably more expensive to generate than its predecessor. Consequently, both versions will continue to coexist, each suited to different research applications.
DeepMind is already plotting the next steps: generating additional datasets for underrepresented cell states and exploring whether future versions could incorporate data across multiple species. The long-term vision includes feeding insights from models like AlphaGenome 2 into virtual cell simulations.
Google is making AlphaGenome 2 available for non-commercial research through its API, with limited access for commercial applications. The release positions DeepMind alongside other AI laboratories that are increasingly training sophisticated systems to tackle major problems in science and medicine.
Discuss this story out loud with our AI speaking coach.
It asks one question at a time, at your level, and gives you feedback at the end.
Start the discussion Retell it3,000+ learners
Also works inChatGPTandClaude
Exercise 3
Discussion
Have a discussion based on the following questions.
- 1.How does understanding cellular context change the way your organization approaches validating genetic findings, and what barriers have you encountered when trying to move from population-level genetics to cellular precision?
- 2.If you were developing a therapeutic program, what cell types would you prioritize studying first to maximize the translational impact of AlphaGenome 2's predictions?
- 3.What existing data gaps in your field do you think AlphaGenome 2 might illuminate, and how would you operationalize using its insights to guide your laboratory experiments?
- 4.How do you think the computational trade-off between AlphaGenome's efficiency and AlphaGenome 2's granularity would play out in your organization's resource allocation decisions?
- 5.Would incorporating multi-species data into future versions of the model fundamentally change how you approach validating findings across biological systems, or would you remain cautious about extrapolation?
Exercise 4
Further discussion
Have a discussion based on the following questions.
- 1.How might AI-driven genomic models like AlphaGenome 2 reshape the traditional structure of academic and industry research workflows, particularly regarding the balance between computational prediction and laboratory validation?
- 2.What ethical and regulatory frameworks need to evolve as AI models become increasingly capable of predicting disease risk and therapeutic response at the cellular level, especially in contexts where predictions may outpace clinical interpretation?
- 3.If virtual cell simulations become sufficiently sophisticated to accurately model drug interactions, how would this fundamentally alter the role of traditional cell culture and animal models in drug development, and what research could no longer be outsourced to computation?
Discuss this story out loud with our AI speaking coach.
It asks one question at a time, at your level, and gives you feedback at the end.
Start the discussion Retell it3,000+ learners
Also works inChatGPTandClaude
Source
This lesson is our own retelling, for English learners, of a story from The Verge.
Related stories

A Mathematician and an AI System Both Claim to Have Solved the Riemann Hypothesis
22 November 2026 · C2

Anthropic's IPO: Can a safety-focused AI company survive shareholder pressure?
22 November 2026 · C1

Self-Teaching AI Systems: A Breakthrough or a Risk?
22 November 2026 · B2

AI and sensors help warn communities about dangerous landslides
22 November 2026 · B1



