Figure 1From: Improving the specificity of exon prediction using comparative genomicsImproving TWINSCAN's prediction on exons. ROC curves by applying the log odds ratio on TWINSCAN's exons. The x-axis is the false prediction rate (FP) of the exon by the log odds score and the y-axis is the true prediction rate (TP) of the exon by the log odds score. The upper graph is the result from human-mouse alignments of TWINSCAN's exons. The lower graph is the result from human-dog alignments of TWINSCAN's exons. The plot shows that by using the log odds score to refine TWINSCAN, we could largely reduce the number of false predictions, e.g., by 32% while keeping over 99% of true positives. The plot also shows that the improvement on TWINSCAN is not affected by the type of alignments used since the two curves are almost identical.Back to article page