Dear Akhil,
clearly you can use many different descriptors in the same analysis, then with usual multidimansional correlation reduction of dimensionality (i.e. by means of PCA) and training/test set procedures you will get a good solution. It is important to stress two basic points that are often overlooked:
1) The need of a real chemically homogeneous series to work with
2) The consciouness 3D descriptors ARE BY NO MEANS SUPERIOR TO 2D ones
(see attached paper)
Dear Akhil,
clearly you can use many different descriptors in the same analysis, then with usual multidimansional correlation reduction of dimensionality (i.e. by means of PCA) and training/test set procedures you will get a good solution. It is important to stress two basic points that are often overlooked:
1) The need of a real chemically homogeneous series to work with
2) The consciouness 3D descriptors ARE BY NO MEANS SUPERIOR TO 2D ones
(see attached paper)
Here is the attached paper I told you in the previous answer, pls note how 2D descriptors outperform 3D, the trick is what we call 2D is not really 2D: structural formulas are in fact very sophisticated symbolic descriptions..
Thanks for your suggestion and related article, great help from your side as i am confused between 2d and 3d. now i am currentlly going to use 2D for my work. one more question as i am new to PCA and dont know much about. In some research article i saw people used GFA for selecting set of descriptors in 2d QSAR.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology