DNA design trick: fit with gradient descent first, then snap to manufacturer catalogue
anshulkundaje · x · 2026-10-04
AlanNawzadAmin explains his DNA sequence design method: discrete optimization over ATCG gets stuck in local minima, so they (1) fit an unconstrained model with gradient descent, then (2) replace each site with the nearest option in the manufacturer's synthesis catalogue. The key observation is that most ATCG distributions are well approximated by catalogue distributions like equal probabilities.
More from Research
- From Text Tokens to Pixels: How Vision Encoders Turn Images Into Meaning — _jaydeepkarale · 2026-10-04
- Paper at NeurIPS: individual parameters in weight-sparse transformers appear interpretable — CatAstro_Piyush · 2026-10-04
- How vector databases work: embeddings, cosine similarity, HNSW, IVF and PQ explained — blaizedsouza · 2026-10-04
- CoFlow adds contrastive trajectory repulsion to Flow Matching, improving few-step generation on ImageNet — burny_tech · 2026-10-04
- ARROW unifies 3D reconstruction and point tracking from arbitrary image sets, sets new SOTA — CSProfKGD · 2026-10-04
- Latent communication papers: KV-cache integrity vs malicious agents and state-delta hybrid channels — burny_tech · 2026-10-04