Fig. 3 | Nature Communications

Fig. 3

From: Biological relevance of computationally predicted pathogenicity of noncoding variants

Fig. 3The alternative text for this image may have been generated using AI.

Performance of tested methods on prioritizing GWAS hits. a Fraction of pathogenic ncSNVs in the top 10 percentile of impact scores declines exponentially as the mixing ratio increases. b AUPRC value decreases as the mixing ratio increases. The gray dotted line presents AUPRC = 0.5 corresponding to random predictions. The black dotted line presents AUPRC = 0.8 corresponding to a desired performance in practice. c Maximum mixing ratio for each tool to achieve AUPRC > 0.8 when pathogenic ncSNVs disrupt ultra-, well-, or least-conserved positions. d Maximum mixing ratio for each tool to achieve AUPRC > 0.8 when pathogenic ncSNVs are inside different types of genomic regions. Since the highest AUPRC of CATO in any category was lower than 0.8, we did not include it in panels c and d. In this task, we used GWAVA unmatched scores

Back to article page