scglue.models.sc.GraphEncoder.forward

GraphEncoder.forward(eidx, enorm, esgn)[source]

Encode graph to vertex latent distribution

Parameters:

  • eidx (torch.Tensor) – Vertex indices of edges (\(2 \times n_{edges}\))

  • enorm (torch.Tensor) – Normalized weight of edges (\(n_{edges}\))

  • esgn (torch.Tensor) – Sign of edges (\(n_{edges}\))

Return type:

torch.distributions.normal.Normal

Returns:

v – Vertex latent distribution (\(n_{vertices} \times n_{features}\))