Seqlet caller gets 22x speedup: 167,750 loci drop from 7.8s to 0.35s
jmschreiber91 · x · 2026-10-21
jmschreiber91 announced a major upgrade to their seqlet caller for genomic attribution analysis:
- extractloci: processing 167,750 loci on one track went from 7.8s to 0.35s with identical inputs (22x faster)
- Dinucleotide shuffling: 64 seqs down from 17.1ms to 2.9ms
- DeepLIFT reference generation: 117ms to 27ms
- New feature: optionally pass a control set of attributions to estimate a threshold for a given FDR level — helps in noisy cases, minimal change when attributions are clean
@eugmattei and @AlMurphy also contributed via DeepLIFT-related reports and PRs.
Related event: AI Auto-Optimization Speeds Genomics Function 22x(3 posts)→
More from Research
- UC Berkeley releases Actor-Gatherer framework decoupling reasoning from context management — adityagp · 2026-10-21
- UW × NVIDIA paper: Compositional RL trains a 27B LLM to be the best teammate in multi-model systems — shangbinfeng · 2026-10-21
- C-RL details: LOO baseline cancels out in GRPO, no extra rollouts needed — shangbinfeng · 2026-10-21
- C-RL-trained 27B model becomes strongest collaborator in 21 of 24 settings — shangbinfeng · 2026-10-21
- Solo RL can make LLMs worse teammates: Shapley-based Compositional RL trains for collaboration — shangbinfeng · 2026-10-21
- Solo RL can make LLMs better alone but worse together, Compositional RL paper finds — shangbinfeng · 2026-10-21