Coding for DNA-based Storage Systems
summary
Built a working rewritable and random-access DNA storage prototype that stores UIUC/Wikipedia fragments, uses G‑block assembly, overextension PCR for editing and Sanger sequencing for verification, and shows zero read errors in the demo at an experimental cost of about $4,500 (synthesis cost per 1,000‑nt fell from ~$2,000 to ~$200 over two years). Model treats synthesis as ~1% substitution noise and sequencing as sampling L‑grams (L≈100) producing a profile vector; codes arise from restricted De Bruijn graphs, Ehrhart counting (LATTE) and Varshamov intersections, with reconstruction via Eulerian paths.
AI-written summary · about