openmolecules.org Forum: Cheminformatics » Clustering by similarity of srtucture

Home » DataWarrior » Cheminformatics » Clustering by similarity of srtucture

Show: Today's Messages :: Polls :: Message Navigator

Clustering by similarity of srtucture [message #2412]

Mon, 21 April 2025 20:55

Rasajna
Messages: 1
Registered: April 2025

Junior Member

Hello, I was using data warrior for clustering molecules from a database based on the structure. I wanted to know more about the parameters and the algorithm being used so I can explain it while presenting my work and just better understand the results. I saw an answer on the forum that said DW uses agglomerative hierarchical clustering but what is the the measure of similarity, is it tanimoto's coeffiecient?

Report message to a moderator

Re: Clustering by similarity of srtucture [message #2413 is a reply to message #2412]

Mon, 21 April 2025 21:11

thomas
Messages: 730
Registered: June 2014

Senior Member

that depends on the descriptor you use. FOr the simple binary fingerprints (FragFp, PathFp, and SphereFp) it is Tanimoto. SkeletonSpheres similarioty is also a kind of Tanimoto, but it is based on fragment count numbers rather than fragment existence bits. OrgFunctions and Flexophore similarities are the result of graph matching algorithms, the first based on best matching trees and the second on an MCS matching of complete graphs. The Flexophore is published and some basic information about the descriptors can be found here: https://openmolecules.org/help/similarity.html.

Report message to a moderator

Previous Topic:

Out of memory

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Thu May 08 07:54:16 CEST 2025

Total time taken to generate the page: 0.03611 seconds