Skip to main content

Table 1 Area under the operating characteristic comparison

From: Quadratic variance models for adaptively preprocessing SELDI-TOF mass spectrometry data

Algorithm/Dataset

PAUC25

PAUC

LibSELDI/HYBRID1

58.8%

(9.9)

66.1%

(7.1)

MassSpecWavelet/HYBRID1

50.8%

(8.5)

64.9%

(5.9)

LibSELDI/HYBRID2

53.2%

(8.7)

64.1%

(6.1)

MassSpecWavelet/HYBRID2

45.4%

(9.5)

61.3%

(6.9)

  1. Area under the operating characteristic curve in a range of false discovery rate values of interest is a useful way to compare peak prediction performance. We show two partial area under the curve metrics, calculated in the range FDR ∈ [0, 50%] (PAUC) and FDR ∈ [0, 25%] (PAUC25). PAUC is more of overall measure of peak prediction potential, while PAUC25 focuses on measuring performance at low FDR. The number shown is the average (standard error) calculated from the 50 operating curves from HYBRID1 and HYBRID2. LibSELDI shows particularly appealing PAUC25 performance.