openmolecules.org

 
Home » DataWarrior » Functionality » Skelspheres descriptor (Skelspheres descriptor)
Skelspheres descriptor [message #1527] Fri, 04 March 2022 14:23 Go to next message
ECG93 is currently offline  ECG93
Messages: 1
Registered: March 2022
Junior Member
It is reported that Skelspheres considers stereochemistry, but how much is the threedimensionality considered? For example when dealing with a library of molecules with high sp3 fraction.

Probably flexophore descriptor would be more appropriate for "Analyse similarity cliffs", but for PCA Skelspheres can't be used as it is not a vector.

Report message to a moderator

Re: Skelspheres descriptor [message #1574 is a reply to message #1527] Sat, 02 April 2022 17:15 Go to previous message
thomas is currently offline  thomas
Messages: 740
Registered: June 2014
Senior Member
The SkelSpheres based similarity is basically a measure of how many circular fragments match between two molecules in regard to both molecules' total count of circular fragments (up to a radius of 5 atoms). Since these fragments largely overlap and since stereo centers within these fragments need to match, a high similarity implicitly goes along with similar 3D-shape of at least a large part of the molecule.

In regard to ligand-to-target binding similarity, the Flexophore is certainly more appropriate than the SkelSpheres, because it allows for scaffold hopping. Even better would be the PheSA shape and pharmacophore similarity, but that requires conformers and is computationally more demanding. Therefore, PheSA similarity is not implemented as standard descriptor and only accessible as 'Superpose conformers...' and as fitness criterion in the evolutionary library creation. The same is true for ligand-to-protein docking.

Report message to a moderator

Previous Topic: Chemistry in 3D
Next Topic: what parameters shall be used to create t-SNE or PCA visualization in Data Warrior
Goto Forum:
  

[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Tue Dec 02 11:40:59 CET 2025

Total time taken to generate the page: 0.19796 seconds
openmolecules.org