DNA sequence models hit SOTA but wet-lab validation remains the bottleneck, experts say
anshulkundaje · x · 2026-09-09
Stanford professor Anshul Kundaje weighed in on Cathy Tie's thread about DNA sequence models: the field has advanced substantially since DeepBind (2015), and current work is close to SOTA — but nowhere near "solved."
Tie's point: 12 years ago Jungla, Deep Genomics and DeepSEA already used ML to predict variants, yet none were clinically actionable because predictions required validation from population data or functional assays. Wet-lab science remains the key to making models useful.
Both agree experimental validation is essential and a real bottleneck; the cited paper includes solid case studies.
More from Research
- Math breakthrough hype hits the New York Times, Harmonic also mentioned — satnam6502 · 2026-09-09
- OpenAI claims Navier-Stokes Millennium Prize solution produced by agent group on next-gen model — AdaptiveAgents · 2026-09-09
- AMS credits Spanish mathematicians Cordoba and Martinez-Zoroa for key Navier-Stokes strategy before OpenAI's final steps — soumitrashukla9 · 2026-09-09
- OpenAI reportedly spent $15M in tokens on Navier-Stokes proof, and it was worth it — bindureddy · 2026-09-09
- AI2 brings papers and talks to ECCV 2026 — allen_ai · 2026-09-09
- MIT TR35 Honoree Shuang Li Uses Language Models and UVA to Make Robots More Adaptable — endernewton · 2026-09-09