Research Article

Comparative Analysis of Mass Spectral Similarity Measures on Peak Alignment for Comprehensive Two-Dimensional Gas Chromatography Mass Spectrometry

Table 3

F1 score and AUC of each peak alignment method for Data II.

CosinePearson’sSpearman’sPartialPart

F1 (%)PAS61.09*61.1749.3161.5858.52
(0.31)#(0.30)(0.25)(0.87)(0.85)
Rank305
DW-PAS59.3259.2356.7459.5258.23
(0.33)(0.34)(0.32)(0.95)(1.02)
151552015
5015
Distance##CCCCC
SW-PAD59.5559.1656.3266.6557.96
(0.41)(0.36)(0.37)(0.77)(0.75)
0.93200.30.70.1
50100
DistanceMhMxMhMhMh
PAM61.4861.5159.4261.7860.19
(0.31)(0.33)(0.36)(0.91)(1.00)
0.050.050.70.10.5
303
DistanceE, Mx, MhMxCE, Mx, MhMh

AUC (%)PAS84.5384.5572.5585.0783.16
DW-PAS83.4583.4381.5083.3082.71
SW-PAD78.0677.8177.7782.9075.35
PAM76.8976.8979.2277.4577.18

Mean (%); standard error (%); the cut-off rank; the rank for the two-step partial and part correlations; the distance measure E, Mx, Mh, and C stand for Euclidean, Maximum, Manhattan, and Canberra distances, respectively; the cut-off similarity score. the weight factor of the mixture similarity score; The numbers in bold and italic indicate the maximum for each of the peak alignment methods.