masashitsubaki / moleculargnn_smiles Goto Github PK

Hello!

I similarly have a question - is it please possible to update the code so that it is possible to also extract the model itself after training?

Thank you and best wishes,
Kadi

Hi,
Nice work! I successfully trained with my own data, but how can I save the model?
Ji

I read your paper and it was very interesting.
I have some questions about your paper and implementation consistency.

The first question is about the numerical formula (3) in your paper.
In your implementation, dose the formula (3) corresponds to update function defined as the aggregation of vectors along the edges in the file of train.py (line 48)?

def update(self, matrix, vectors, layer):
        hidden_vectors = torch.relu(self.W_fingerprint[layer](vectors))
        return hidden_vectors + torch.matmul(matrix, hidden_vectors)

If then, sigmoid function is not used in line of the return process, though the formula (3) describes sigmoid.

The second question is about formula (5, 6) in your paper.
The update of the edge information through the activation function is described in this formula (5, 6), but this process dose not seems to be implemented. Edge information is only used for extraction fingerprint based on r-radius subgraph.

Is the implementation of molecularGNN in this repository a simpler version of your paper?

Thank you

Where can I get the original datasets ('photovoltaic' in regression and 'HIV' in classification) used in the codes.
I am interested in the databases of used datasets. I'll appriciate it if the websites are provided.
Many thanks!

Hi,
Your dataset has 2 labels: 0,1. My dataset has 185 labels [0,1..,184]. When i run your code with my dataset i got errors. which part of code should i change for 185labels? Thanks.

I'm trying to understand your code but I have a few doubts.

First of all I'd like to understand how you generate the r-radius subgraphs. For instance, if I run your code, the first SMILES is 'CCC1=[O+][Cu-3]2([O+]=C(CC)C1)[O+]=C(CC)CC(CC)=[O+]2', which corresponds to the molecule vector (or fingerprint) [ 6, 7, 8, 9, 10, 9, 8, 7, 6, 11, 9, 8, 7, 6, 11, 8, 7, 6, 9, 12, 12, 12, 13, 13, 13, 13, 12, 12, 12, 13, 13, 13, 13, 12, 12, 12, 13, 13, 13, 13, 12, 12, 12]. I'm not able to understand how to obtain this vector. I though you were just looking for all the sub-graphs within a certain radius from each atom of the molecule. In this case I would expect something like:
[0, 1, 2] [0, 1, 2, 3, 9] [0, 1, 2, 3, 4, 6, 9] [1, 2, 3, 4, 5, 9, 10, 18] [2, 3, 4, 5, 6, 10, 11, 15, 18] [3, 4, 5, 6, 7, 9, 10, 18] [2, 4, 5, 6, 7, 8, 9] [5, 6, 7, 8, 9] [8, 6, 7] [1, 2, 3, 5, 6, 7, 9] [3, 4, 5, 10, 11, 12, 14, 18] [4, 10, 11, 12, 13, 14, 15] [10, 11, 12, 13, 14] [11, 12, 13] [10, 11, 12, 14, 15, 16, 18] [4, 11, 14, 15, 16, 17, 18] [14, 15, 16, 17, 18] [16, 17, 15] [3, 4, 5, 10, 14, 15, 16, 18]

Moreover, I do not understand why the second entry in molecules does not start at zero anymore: [20, 21, 22, 23, 24, 24, 24, 23, 25, 9, 10, 9, 25, 20, 21, 22, 23, 24, 24, 24, 23, 26, 27, 28, 23, 24, 24, 24, 23, 29, 29, 27, 28, 23, 24, 24, 24, 23, 26, 30, 31, 32, 32, 32, 32, 32, 30, 31, 32, 32, 32, 32, 32, 13, 13, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 13, 13].

Finally, I'd like to understand if you concatenate all the r-radius subgraphs into a single vector (one of the vectors above). Thank you.