AuraTracer智迹闻
中文

EVENT DOSSIER

Google DeepMind Maps 9 Billion Possible DNA Variants

DeepMind released AlphaGenome Atlas on September 8, 2026, and made it open source.

2026-09-08 22:00 Science 🔥 59.2 heat score hn #1techmeme #4
2sources
1days unfolding
59.2heat score
7mentions
SummaryAI generated

On September 8, 2026, Google DeepMind released the genome mapping tool called AlphaGenome Atlas, aimed at addressing the fundamental research challenge of how DNA variations affect gene regulation. Based on the AlphaGenome AI model, this tool provides predictions for 9 billion single-letter variations and over 100 million short insertions or deletions in approximately 30 billion base pairs of the genome, with a total data volume of 1 PB. Atlas includes a new feature called Single Value Impact Score (AVI), which helps scientists quickly assess the significance of variations and select priority subjects for verification, thereby accelerating disease research and treatment development. Currently, the tool is available online in a repository for free use by non-commercial researchers; although some researchers still face operational barriers due to the need to write code, approximately 9,000 researchers have already used the tool through API calls.

Related eventsRELATED EVENTS
Quick factsQUICK FACTS
9 billionNumber of variations
More than 100 millionNumber of short insertion or deletion variations
1PBTotal data volume
Key entitiesKEY ENTITIES
AlphaGenomeDeepMindGoogleGoogle DeepMindNaturePushmeet KohliUniversity of British Columbia

Event frameEVENT FRAME

Launch

Google DeepMind AlphaGenome Atlas 发布包含90亿种变异预测的Atlas

Status

DeepMind released AlphaGenome Atlas on September 8, 2026, and made it open source.

Coverage · reports per dayLANGUAGE SPLIT

Entity relations
AlphaGenome × Nature2AlphaGenome × Google De…1AlphaGenome × Pushmeet …1AlphaGenome × Universit…1Google DeepMind × Nature1Google DeepMind × Pushm…1

Integrated timelineUNIFIED TIMELINE

  1. 2026-09-08

    DeepMind released AlphaGenome Atlas

    Google DeepMind released AlphaGenome Atlas on September 8, based on the AlphaGenome AI model, predicting the impact of 9 billion single-letter variations and more than 100 million short insertion or deletion variations in approximately 30 billion base pairs of the genome. Total data…

    2 reports

SignalsSIGNALS

Keyword heat
  • AlphaGenome2
  • Nature2
  • Google DeepMind1
  • University of British Columbia1
  • Pushmeet Kohli1
  • DeepMind1
  • Google1

All reports (2)SOURCES

N Nature News en 2026-09-08 08:00

DeepMind’s new genome ‘atlas’ charts effects of all 9 billion human gene mutations

Google DeepMind 发布新基因组“图谱”,利用 AlphaGenome AI 模型预测人类基因组中所有 90 亿种单核苷酸突变的影响。该图谱还收录了超过 1 亿个短插入或缺失变异,旨在辅助诊断罕见病、揭示常见疾病隐藏机制及 DNA 调控规则。图谱数据总量达 1 拍字节,由 DeepMind 团队开发用于量化变体生物学影响的 AlphaGenome Variant Impact (AVI) 评分指标提供。目前该工具免费供非商业用途使用,虽已吸引约 9000 名研究人员通过 API 调用,但部分生物学家仍因需编写代码而面临门槛。

I IEEE Spectrum AI en 2026-09-08 22:00

Google DeepMind Maps 9 Billion Possible DNA Variants

Google DeepMind 于 9 月 8 日发布 AlphaGenome Atlas,提供预先计算好的全基因组 90 亿种单字母变异预测。该工具基于 AlphaGenome 模型,旨在解决理解 DNA 变异如何影响基因调控这一基础疾病研究难题。Atlas 以在线仓库形式公开,供非商业研究免费使用,无需用户编写代码或运行模型,并新增单一数值影响评分功能以快速评估变异意义。尽管 AlphaGenome 仅分析约 100 万碱基对区域,难以预测远距离增强子等复杂调控效应,但 Atlas 可帮助科学家筛选可能性并优先验证实验,从而加速基础生物学、疾病研究及治疗开发。数据集包含全基因组约 30 亿碱基对的 90 亿种变异,总量约 1PB,其计算速度较早期预估提升了 80 倍。