Table 1 The cosine similarity between predicted and the ground truth mass spectra. MoMS-Net shows the best performance on both the FT-CID and FT-HCD datasets.

From: Mass spectra prediction with structural motif-based graph neural networks

 

FT-CID

FT-HCD

CNN

\(0.356\pm 0.002\)

\(0.535\pm 0.002 \)

MassFormer

\(0.385\pm 0.005\)

\(0.573\pm 0.003 \)

WLN

\(0.357\pm 0.001\)

\(0.569\pm 0.001 \)

GCN

\(0.356\pm 0.001\)

\(0.565\pm 0.001 \)

MoMS-Net (Ours)

\(\mathbf {0.388\pm 0.002}\)

\(\mathbf {0.578\pm 0.001 }\)

  1. Every result is repeated five times with different random seeds. Significant values are in bold.
Back to article page