Supplements of the following paper:
Yamanishi Y., Bach F. and Vert J.-P., Glycan classification with tree kernels
Data for glycans
Class labels of bloods (365 glycans)
Note that each row is as follows:
[Glycan ID], [each blood component or not].........
- Plasma
- Serum
- Leukemia
- Erythrocyte
Features of co-rooted subtrees for all possible combination of D and N
- Features (d=no,0,1,2,3,4,5, N=1,2,3,4.....17,18,all)
dn means we donot consider the depth information of the root of each subtree.
d0 means we use the depth information of the root of each subtree.
d1 means we shift the depth information by 1.
d2 means we shift the depth information by 2.
......
d5 means we shift the depth information by 5.
n1 means we use a set of subtrees of size 1 (number of nodes is 1).
n2 means we use a set of subtrees of size 2 (number of nodes is 2).
......
n18 means we use a set of subtrees of size 18 (number of nodes is 18).
n0 means we use a set of all subtrees.
e=0 means that we donot remove the external edge labels.
e=1 means that we remove the external edge labels.
Kernels of co-rooted subtrees for all possible combination of D and N (spectrum kernel)
- Kernels (d=no,0,1,2,3,4,5, N=1,2,3,4.....17,18,all)
Note that each row is as follows:
[Glycan ID], [kernel values1] [kernel values2].........
The first colum corresponds to the glycan ID.
Features of all possible subtrees for all possible combination of D and N
- NEW Features (d=no,0,1,2,3,4,5, N=1,2,3,4.....17,18,all)
Kernels of all possible subtrees for all possible combination of D and N (spectrum kernel)
- NEW Kernels (d=no,0,1,2,3,4,5, N=1,2,3,4.....17,18,all)
Note that each row is as follows:
[Glycan ID], [kernel values1] [kernel values2].........
The first colum corresponds to the glycan ID.