psearch functionality

[spadavec]

Hello

I see a reference to a "psearch" functionality for high throughput pharmacophore screening, but can't seem to find it anywhere. Can that be shared? If not, what are the best ways to screen a large collection of molecules (1M+) against a few point pharmacophore?

[DrrDom]

Hi Vito,

there are few options:

1. Use our psearch software from this branch of the repo https://github.com/meddwl/psearch/tree/dev2. You will need to generate a database of conformers and pharmacophores for each molecule. The drawback of our approach that some molecules can be missed due to fuzzy matching. However, the fewer points in a query will be the less number of compounds will be missed.
2. Use pharmit software of David Koes - https://sourceforge.net/p/pharmit/code/ci/master/tree/src/. You will still need to generate a database of conformers and pharmacophores. The generation of a database may take longer and it can be larger, but screening will be very fast. So, if you will be able to install it, I would recommend it for screening. It uses true 3D alignment.

I do not know what would be more effective for a pharmacophore with a small number of features. How many features are in your pharmacophore?

Pavel.