Attentive FP with DGL-LifeSci #RDKit #DGL #Chemoinformatics

I posted some code about DGL before. Recently DGL team published life science focused package DGL life sci. It seems almost as same as DGL but code is well organized. User can install the package from conda!

And I got some comments about my attentive FP example.

So I tried to use new DGL-LS and build attentive fp model.

Now 11:10PM in Japan so I’m so sleepy… So I just pasted my jupyter notebook and go to bed Zzzzzzzz.

Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.

https://nbviewer.jupyter.org/gist/iwatobipen/017fae146c1d67a2cfee455c0a427330

I added color bar in drawing code.

The code always normalize atom weight and coloring it so I think range of weight is always 0 – 1.

I felt that the package will be good starting point of chemoinformatics. And some new functions are implemented. I’ll check and play them.

Good night!

Advertisement

Published by iwatobipen

I'm medicinal chemist in mid size of pharmaceutical company. I love chemoinfo, cording, organic synthesis, my family.

7 thoughts on “Attentive FP with DGL-LifeSci #RDKit #DGL #Chemoinformatics

  1. Thanks iwatobipen! It is a very clear demonstration on how to implement attentiveFP for classification task and visualize its prediction results!

    Really appreciate your effort!

      1. Hi iwatobipen,
        You used self-defined atom_featurizer and bond_featurizer and mol_to_bigraph in the script.

        Can explain why you used self-defined featurizer instead of the default featurizer e.g. CanonicalAtomFeaturizer() in dgl and mol_to_bigraph instead of mol_to_complete_graph?

        Dose the featurizer and graph method giving better results?

        Many thanks in advance!

      2. Hi Mark,
        Hope you are doing well. And thanks for your question.
        It’s short answer I defined the feature because I would like to use same features as original publication.
        https://docs.dgl.ai/en/0.4.x/generated/dgl.data.chem.CanonicalAtomFeaturizer.html
        DGL default featurizer has different features to the article.
        https://docs.dgl.ai/en/0.4.x/generated/dgl.data.chem.CanonicalAtomFeaturizer.html
        And I haven’t tried AttentiveFP with different featurizer. Hope this will answer for your question.
        Thanks,

      1. Thanks a lot for your kind comments,I’m looking forward to your reply.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: