datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
clinvar-diploid-snv
ClinVar Diploid SNV — mode-of-inheritance-aware genotype benchmark
A zygosity-sensitive relabelling of high-confidence ClinVar SNVs. Each variant is instantiated in
both a heterozygous and a homozygous state, and the label is assigned at the genotype level
under simplified dominant/recessive inheritance rules. A pathogenic recessive variant is therefore
negative as a heterozygous carrier and positive as a homozygote — so variant identity alone is
insufficient to predict the… See the full description on the dataset page: https://huggingface.co/datasets/scrc-dnai/clinvar-diploid-snv.clinvar-diploid-indel
ClinVar Diploid Indel — zero-shot genotype benchmark
High-confidence ClinVar insertions and deletions, each instantiated in a heterozygous and a
homozygous state and labelled at the genotype level under the same simplified inheritance rules as
the SNV benchmark.
Its purpose is narrower: to test whether a representation that keeps the actual inserted and
deleted bases — rather than collapsing every indel into one generic symbol — carries usable
variant-effect signal. Evaluated… See the full description on the dataset page: https://huggingface.co/datasets/scrc-dnai/clinvar-diploid-indel.