Additional file 3.

Supplementary figure 2. Predicted distributions of identifiable and unidentifiable for a sample from the UCD test dataset (A) and sample A1 from the ISB dataset (B). The number of spectra is calculated with spectra placed in bins of 0.25 for the discriminant score. Spectra with fewer than five peaks are removed in line with the EM-algorithm. The solid line represents the actual distribution of spectra of the complete dataset and the dotted line represents the estimated distributions of identifiable and unidentifiable spectra using the expectation-maximization algorithm.

Format: TIFF Size: 516KB Download file

Wong et al. BMC Bioinformatics 2007 8:51   doi:10.1186/1471-2105-8-51