. 2022 Mar 3;38(9):2544–2553. doi: 10.1093/bioinformatics/btac133

Table 1.

Existing approaches that we consider and their generalized NoN counterparts

Single-level approaches		NoN approaches
Level 1	Level 2	Level 1 + Level 2
L1 GDVM (73 $n_{L 1}$ )	L2 GDV (73)	L1 GDVM + L2 GDV
L1 GCM (121)	L2 GDV (73)	L1 GCM + L2 GDV
L1 DiffPool (64)	L2 SIGN ( $n_{L 2}$ )	L1 DiffPool + L2 SIGN
		Combined all¹
		Combined all²

Note: The value in parentheses next to each Level 1 approach indicates the size of each Level 1 network’s feature. For features that are matrices, the rows are concatenated to form a 1D feature. $n_{L 1}$ is the size of the Level 1 network for which the feature is being computed. The value in parentheses next to each Level 2 approach indicates the size of each Level 2 node’s feature. $n_{L 2}$ is the number of nodes in the Level 2 network. Sizes of NoN features (not shown) are sums of their corresponding single-level approaches, as the NoN features are formed via concatenation. ‘Combined all¹’ refers to L1 GDVM + L2 GDV + L1 DiffPool + L2 SIGN, which is used for synthetic NoNs ‘Combined all²’ refers to L1 GCM + L2 GDV + L1 DiffPool + L2 SIGN, which is used for the biological NoN.