Pytorch-Geometric graph classification where graphs all have the same shape

Hi all,
as the headline suggests, I want to train a model for graph classification using Pytorch-geometric.
As a matter of fact, I actually have many samples based on the same graph structure, differing in their node features (one node-feature vector per sample).

I tried to create a data object for each sample (referring to it as a different presentation of the same graph), but that is very computationally expensive, because there are relatively many such samples.

Therefore, I’m looking for a way to do graph-level classification, with a batching system that will refer to all graphs as having the same nodes, and the same number of node features.

Is that possible?