Table 1 Construction of training and test sets of glycopeptide spectra matches (GSMs) of N-glycopeptides identified from IgG and AGP standards and their classification of fucosylation types both manually and by machine learning methods such as the support vector machine (SVM) and deep neural network (DNN).
From: Machine Learning Classifies Core and Outer Fucosylation of N-Glycoproteins Using Mass Spectrometry
N-glycoproteins (IgG & AGP Standards) | Training Set (433 GSMs) | Test Set (393 GSMs) | |||||||
|---|---|---|---|---|---|---|---|---|---|
None (%) | Core (%) | Outer (%) | Dual (%) | None (%) | Core (%) | Outer (%) | Dual (%) | ||
Classification Methods | Manual Classification | 170 (39.2%) | 106 (24.5%) | 89 (20.5%) | 68* (15.7%) | 162 (41.2%) | 70 (17.8%) | 96 (24.4%) | 65* (16.5%) |
SVM Classification | 170 (39.2%) | 106 (24.5%) | 89 (20.5%) | 68 (15.7%) | 163 (41.5%) | 70 (17.8%) | 97 (24.9%) | 62 (15.8%) | |
DNN Classification | 170 (39.2%) | 106 (24.5%) | 89 (20.5%) | 68 (15.7%) | 163 (41.5%) | 70 (17.8%) | 98 (24.9%) | 62 (15.8%) | |
