Can the dataset of graph classification task be used for node classification training?

[jiaruHithub]

Dear developers:
I want to know: Can the dataset of graph classification task be used for node classification training?
Such as PROTEINS and NCI1.
If this is possible, where are the examples?
Thank you for your answer!

[rusty1s]

A graph classification dataset contains only a single label per graph, while in node classification you have a label for each node. If you want to cast a graph classification dataset to a node classification one, I think it's reasonable to set each node label to the ground-truth label of the given class, e.g.:

def transform(data):
    data.y = data.y.repeat(data.num_nodes)
    return data

dataset = TUDataset(..., transform=transform)

[jiaruHithub]

Thank you!