Find the best descriptor combination

KNIME Analytics Platform
#1

Dear community,

I have the following problem:

I have a given set of molecule and a given set of reference molecule which are known highly potent substances. After calculating certain descriptors I want to compare the molecules with the reference set by a distance similiarity search and my aim is to minimize the distance. For this purpose trying different descriptor combinations is necessary. So I have for example 20 rows each containing a different descriptor.
Is there any simple way or a node which offers the possbility to try every combination of the calculated descriptors?

I hope you understand my problem.
Thank you for your help and best regards,

Giuseppe

#2

Not really an answer but you could look at the Similarity Search node. Yes, you can also calculate distances with that.

To add to this, for most distance based metric you will also need to look at the normalize node as the descriptors must be normalized for example for Euclidean distance to work correctly.

#3

Sry I meant that the each COLUMN contains a calculated descriptor.

Yes, I already found the Similarity Search Node. Now I just want to solve the problem to try all possible combinations.