Saturation genome editing

Predict continuous functional damage for assayed SNVs in endogenous-locus saturation genome editing screens, using the gene, assay mechanism, and local exon sequence.

Task design

published gene panels, each containing 50 variants from one exon and exactly 100 unmarked flanking bases on each side. Panels prefer 25 missense and 25 local splicing variants and sample across measured-damage quantiles.

The primary score is mean within-gene Spearman correlation. Mean Pearson correlation reports numerical agreement, valid-output rate reports strict JSON compliance, and invalid completed outputs contribute zero while remaining identifiable as format failures.

Task version: 1.0 · question schema 2.0
Output: FINAL: {"V01": number, ...}
Questions: Public development set

Interpretation

Spearman measures ordering within each gene, not cross-assay calibration. SGE effects depend on the cellular system, engineered background, selection, timing, and treatment; they are not clinical classifications. One exon window omits distant gene and splice context, while class and quantile sampling do not reproduce the natural variant distribution.

Questions

Inspect the exact prompt given to a model alongside its complete response.

questions match the current filters · select one row to inspect it