Reading these numbers

The model is LD score regression with the Evo2 score as the single continuous annotation (1000G EUR, MAF ≥ 5% reference). No baseline-LD annotations are included, so the coefficient measures the annotation's association with per-SNP heritability on its own — not its contribution over and above conservation, coding status and the other baseline-LD properties. The interpretation depends on the selected annotation: signed Δ describes a likelihood-shift direction, while |Δ| describes its magnitude. These results do not establish added information beyond conservation.

The coefficient is per unit of the annotation, so its absolute size carries no meaning on its own; it is given with its standard error and 95% confidence interval, and the z-score is the quantity to compare. Because the scale is not comparable across annotations or cohorts, the rank and percentile within each biobank are shown alongside.

h² is the trait's total observed-scale SNP heritability from the same run: a large z on a trait with h² near zero is partitioning almost nothing. The LDSC intercept helps assess inflation beyond that attributed to polygenic signal; interpret it alongside the study's design and quality-control information.

Each biobank is ranked only against its own traits — the three sets differ in phenotype definition, ancestry and sample size, so z-scores are compared as ranks across biobanks, never pooled. Search matches are candidate phenotype records, not verified matches of the same study or evidence of replication. A search with no matches does not establish that the phenotype has no stored result under another original name.