Add Martin 'A Bioavailability Score'

[ikmckenz]

From the paper:

A Bioavailability Score, ABS, is formulated as the probability that a compound will have >10% bio-availability in rat or measurable Caco-2 permeability. ABS is 0.11 for anions for which PSA is >150 Å2, 0.56 if PSA is between 75 and 150 Å2, and 0.85 if PSA is <75 Å2. For the remaining compounds ABS is 0.55 if it passes the rule-of-five and 0.17 if it fails.

[jrytved]

I'd give this a go, but I'm stalling a bit on how we would determine whether a compound is an anion or not. If a formal charge is specified in the SMILES input, it would be straightforward. For a novel compound i don't see a straightforward way to predict a pKa-value and calculate the charge in a range around physiological pH. Any thoughts on this? I would love to help out on this project.

[ikmckenz]

SMILES does contain charge information, and we can easily compute the formal charge using RDKit, but I don't know a good way to predict a pKa-value of the predominant form of a molecule in a specific PH.

[fernan]

Predicting pKa values computationally is not straightforward (see DOI:10.1021/acsmedchemlett.1c00435). In any case, it would require calling an external library or program from adme-pred.py. Haven't done a thorough search, and have not tested it myself, just wanted to drop Opera here (https://github.com/NIEHS/OPERA), as one of the few open /free predictor.

…

On Fri, Mar 31, 2023 at 2:50 PM Ian McKenzie ***@***.***> wrote: SMILES does contain charge information, and we can easily compute the formal charge using RDKit, but I don't know a good way to predict a pKa-value of the predominant form of a molecule in a specific PH. — Reply to this email directly, view it on GitHub <#6 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AABK243LQIKIYFKR36JYYEDW64KMZANCNFSM4LJXLZIQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- fernan