CRM Prediction Accuracy on a Dataset with 208 “Positive” Sequences (CRMs) and 208 “Negative” Sequences (Random Noncoding Genomic Fragments)

<div><p>The <i>y</i>-axis shows the number of negatives included in a given number (<i>x</i>-axis) of top scoring sequences. Green line (diagonal) represents error rate expected by chance.</p><p>(A) Predictions were made based on percentage identity (PID, purple line), Stubb score (blue), Morph LLR1 score (red), and Morph LLR2 score (black).</p><p>(B) Predictions were based on the total number of binding sites predicted by Stubb (blue) and Morph (black).</p>