Skip to main content
Figure 10 | BMC Bioinformatics

Figure 10

From: Building blocks for automated elucidation of metabolites: Machine learning methods for NMR prediction

Figure 10

Representative examples for typical misclassifications ('hard errors'). NMRShiftDB-molid: 10019871, atomID: 10189925 at heavy atom nr. 16 (0.62 ppm), atomID: 10189928 at heavy atom nr. 17 (0.27 ppm). NMRShiftDB-molid: 10021849, atomID: 10324987 at heavy atom nr. 22 (0.73 ppm), atomID: 10324990 at heavy atom nr. 23 (0.27 ppm). NMRShiftDB-molid: 20062578, atomID: 20958256 at heavy atom nr. 17 (3.53 ppm), atomID: 20958257 at heavy atom nr. 17 (2.72 ppm). NMRShiftDB-molid: 10027587, atomID: 11110950 at heavy atom nr. 7 (2.11 ppm), atomID: 11110951 at heavy atom nr. 7 (1.45 ppm). NMRShiftDB-molid: 10021815, atomID: 10323000 at heavy atom nr. 13 (1.96 ppm), atomID: 10323001 at heavy atom nr. 13 (1.11 ppm). NMRShiftDB-molid: 10022075, atomID: 10335470 at heavy atom nr. 13 (3.61 ppm), atomID: 10335471 at heavy atom nr. 13 (2.90 ppm). NMRShiftDB-molid: 21156, atomID: 283794 at heavy atom nr. 1 (8.21 ppm). NMRShiftDB-molid: 21324, atomID: 296084 at heavy atom nr. 2 (6.73 ppm). Provided shift values are experimental and not predicted shifts. Predictions are given in table 2. The corresponding descriptor set is provided in table 2 of additional file 1. The HOSE codes of these atoms are shown in table 4. The different descriptor sets for atoms 10189925 and 10189928 are shown in table 3.

Back to article page