DrugEncoder.AttentiveFP

Click here to go back to the reference.

class AttentiveFP(nn.Module):
    def __init__(self, x_num_embedding: int = _x_num_embedding, edge_num_embedding: int = _edge_num_embedding,
                 embedding_dim: int = _drug_dim, ft_dim: int = _drug_dim, dropout: float = _dropout,
                 hid_dim: int = 384, num_layers: int = 2, num_steps: int = 3):
        """hid_dim, num_layers, num_steps"""
        super(AttentiveFP, self).__init__()
        assert num_layers >= 1
        self.x_embedding = nn.Embedding(x_num_embedding, embedding_dim)
        self.edge_embedding = nn.Embedding(edge_num_embedding, embedding_dim)
        self.reset_parameters()
        self.encode_AttentiveFP = models.AttentiveFP(in_channels=embedding_dim, hidden_channels=hid_dim,
                                                     out_channels=ft_dim, edge_dim=embedding_dim, num_layers=num_layers,
                                                     num_timesteps=num_steps, dropout=dropout)

    def reset_parameters(self):
        torch.nn.init.xavier_uniform_(self.x_embedding.weight.data)
        torch.nn.init.xavier_uniform_(self.edge_embedding.weight.data)

    def forward(self, g):
        x, edge_index, edge_attr, batch = g.x, g.edge_index, g.edge_attr, g.batch
        x = self.x_embedding(x).sum(1)
        edge_attr = self.edge_embedding(edge_attr).sum(1)
        x = self.encode_AttentiveFP(x, edge_index, edge_attr, batch)
        return x