Figure 9

Calibration curve for estimating annotation precisions from MW scores. The x axis denotes MW scores; the y axis denotes annotation precision. The predictions are binned into 50 equally spaced intervals based on associated MW scores. Annotation precision for each bin is calculated based on the comparisons of the predicted annotations and the original annotations for known proteins. The purple dots denote the binning results. The red curve shows curve fitting results with a two parameter logistic regression function, y = exp(a + b*x)/(1+ exp(a + b*x)). a = -6.71 and b = 13.57 (with 95% confidence bounds).