It will take 3 years after changing job. Fortunately, I’m still working at early drug discovery field as chemoinformatician ;)
As many drug hunter know that bioisosteric replacement is one of the major way for designing novel molecules from known biologically active molecules.
Transformations from carboxylic acid to tetorazole, from amide to oxazole are major examples of bioisosteric replatecement.
Are these pair, “carboxylic & tetorazole” and “amide & oxazole” simlar ? And are there any metrics to compre these similarity instead of biological activities?
Espsim is one of the interesting OSS package for compare molecules with electro static potentials.
I tried to compare following tetrazole and carboxylic acid pair and amide and oxazole pair.
To clarify the similarity, I used not only neutoral state of carboxylic acid but also ionization state.
Let’s compare carboxylic acid and tetrazole with Morgan FP and ESPsim. As you can see that tetrazole and carboxylic acid shows low 2d finger print similarity but high similarity with ESPsim and Shapesim.
Next, let’s compare carboxylate and carboxylic acid metyl ester.
Interestingly ESPSIM between calboxylate and methyl ester is very low compared to their shape similarity.
How about similarity between calboxylic acid and methyl ester? Ah! there molecule shows high similarity not only shape but also electrostatic potential.
Next, let’s see amide and oxazole.
The result is as same as calboxylate case, the pair shows low 2D similarty but high shape and esp similarity.
In this experiments, ESPsim is interesting tool for predicting bioisosteric fragments. I think it worth to use espsim as a tool for drug design.
Today’s code is uploaded my gist.
This is the last chemoinformatics blogpost in this year. Thanks for visiting and reading my blog. I hope you and your family have a great rest of the year and have a happy new year ;)
Is life worth living? 