tangermeme encodes all 455 hg38 contigs in 1.5 seconds after auto-optimize speedups
jmschreiber91 · x · 2026-10-11
jmschrei reports a new round of tangermeme optimizations: onehotencode, tuned via an auto-optimize search, cut geometric-mean latency from 0.1204 ms to 0.02729 ms and can now encode all 455 contigs of hg38 in 1.5 seconds. The optimizations include caching the 256-entry byte lookup table, writing into arrays without zeroing first, encoding two bases per lookup on long sequences, and a new njobs parameter, with outputs verified identical to main across 1,823 tests. The author notes these very fast primitives, combined with agentic skills, let agents work much faster on genomics tasks.
More from coding & agent
- Every Agent Works in Slack: Tag It, Assign a Task, and It Learns From Corrections — every · 2026-10-11
- Developer Builds Interactive 3D SpaceX Raptor 3 Model with Opus 5.5 and Three.js — techartist_ · 2026-10-11
- Developer Adds SteveCAD MCP Server to His Engineering Agent Wright — burhop · 2026-10-11
- GPT Researcher crosses 30k GitHub stars, built and maintained almost entirely by community — EdenEmarco177 · 2026-10-11
- shadcn: AI hasn't sped up abstraction design yet, it just helps me fail faster — shadcn · 2026-10-11
- Free 7-week production Agentic RAG course hits 9.7k stars on GitHub — mdancho84 · 2026-10-11