Variant effect predictors are shaping the ClinVar labels used to train and evaluate them
pastramimachine · x · 2026-10-28
A shared research link highlights a circularity problem: outputs of variant effect predictors are influencing ClinVar pathogenicity classifications, which in turn are used to train and evaluate those same predictors — a cautionary case for benchmark data quality in AI-for-science.
More from Research
- Compute for software correctness will flip: verification over fuzzing — moarbugs · 2026-10-28
- DeepMind open-sources HOPE, a math framework for quantifying neuron capacity — teortaxesTex · 2026-10-28
- Token embeddings matter far more to value vectors than norms suggest — teortaxesTex · 2026-10-28
- Tweak one neuron and Qwen3.5-4B turns into a screaming oracle: PEFT activation steering demos — LysandreJik · 2026-10-28
- LiveBench: a contamination-limited LLM benchmark with fresh questions and objective scoring — QuixiAI · 2026-10-28
- Meta Models predict LLM training loss curves before runs finish — ZimingLiu11 · 2026-10-28