theochem / b3db Goto Github PK

Hi,

I've noticed that there's a smiles_result column in the regression dataset, which is not present in the classification dataset.

What exactly is this, is this the cleaned up SMILES or something else? If it is, why is it not present in the classification dataset?

Thanks

A license for this database is required.

https://github.com/theochem/B3DB/security/dependabot/2

Add About and keywords in Github.

There are some nan strings in the final data and should be removed.

Problem Statement: Having a simple, no-nonsense queried database of antivirals that are BBB permeable or not?

Preliminary Approach: @fwmeng88 's upcoming computational model to be applied to a comprehensive list of antivirals taken up from https://www.viprbrc.org/brc/home.spg?decorator=vipr and then building a simple database to show the BBB permeability.

Outcomes: Since I don't see similar results after an initial literature review, this represents a low hanging fruit, which might be useful to the community. It is good to have data on antiviral BBB permeability for training future models. Though people can grab a list of antiviral SMILES from PubChem, and run them through SwissADME for BBB permeability data, and a visual inspection through EggPlot: https://chemistry-europe.onlinelibrary.wiley.com/doi/10.1002/cmdc.201600182. This effort might compare the two models, helping find outliers from the SwissADME data.

I hope this is interesting enough to spend some time, and develop it together. Happy to stay in the loop, and chat.

Hi, I would like to know whether the molecules with categorical labels are experimentally determined or computationally predicted. Thank you

There are more molecules with BBB permeability information that should be added in the next release. @fwmeng88

We should add the extra molecules from B3cls to B3DB when we finish that work.

The molecule 847 - 3-methylhexane (SMILES C=C([CH]CC)CC) is problematic as there is a carbon atom with unusual valence and the SMILES does not match SMILES for 3-methylhexane. This should be https://pubchem.ncbi.nlm.nih.gov/compound/11507.

Credit goes to Dr. Andrea Mauri from Alvascience.