Award Winner
« Back
Bhaskar Balaji wins Best Poster Award at the Drug Discovery India 2014 conference
Bhaskar Balaji wins Best Poster Award at the Drug Discovery India 2014 conference

The poster discusses the important theoretically calculated descriptors that have relationship with the inhibition of 17ß-HSD2 which can contribute to the better design of 17ß-HSD2 inhibitors.
Poster
QSAR prediction of 17β-hydroxy steroid dehydrogenase 2 inhibitors using support vector machine
QSAR prediction of 17β-hydroxy steroid dehydrogenase 2 inhibitors using support vector machine
Bhaskar Balaji, Muthiah Ramanathan
Department of Pharmacology, PSG College of Pharmacy
Poster Views: 3,511
view poster »
Poster Abstract
Objectives
To find out valuable descriptors from potent and diverse chemical class molecules.
To depict the difference between linear and non linear statistic model in predicting the descriptors.

Methodology

Dataset Collection

311 molecules were taken for this study belongs from diverse scaffolds of hydroxy phenyl naphthol (Wetzel et al., 2011), triazole (Henn et al., 2012), pyrolidinones, biphenyls, bicyclic substituted hydroxy phenyl methanones (Wetzel et al., 2012) and thiophene amides (Oberwinkler et al., 2013).

Descriptor generation

Molecular descriptors were generated by QikProp, Schrodinger and PaDEL. Step wise multiple linear regression and principal component analysis was employed to find out relevant descriptors.

Support vector machine

Libsvm package was used to perform SVM analysis.
Radial basis function kernel was adopted for SVC for solving multiclass problem.
This method constructs k(k-1)/2 hyperplane which was built using the training data chosen out of k class (k=3).
For method validation, accuracy and R2cv (LOO cross validation) was measured.

Results and Conclusion

SVM model found better informative than MLR exhibiting non linear relation between biological activity and molecular descriptors.

Existing SVR model can be utilized to reconstruct the molecular scaffolds and structure for better biological activity.





Related Posters


Evaluation of the Orion Outlier Detection Algorithm for Data Streams
Nam Phung

A Novel Method For Discovery of Peripheral Blood Biomarkers in Idiopathic Pulmonary Fibrosis Using Extensive Depletion and TMTcalibratorTM Tissue-Enhanced Plasma Proteomics
I. Pike1, M. Bremang1, P.J. Wolters 2, R. Gaster3, S. Turner3, M. Decaris3

Addressing Technical, Legal and Diplomatic Obstacles To Combatting Human Trafficking Using DNA Databases Across National Boundaries
Michael J. Hennesseya, Jon Krollb, Jose A. Lorentec and Howard D. Cashb*

The EurOPDX EDIReX project: towards a European Research Infrastructure on patient-derived cancer models
E. Vinolo 1, J.P. Morris 1, D.G. Alférez 2, J. Arribas 3,4,5, C. Bernadó 3,4,5, A. Bertotti 6, A. Bruna 7, A.T. Byrne 8, C. Caldas 7, R.B. Clarke 2, N. Conte 9, R. Corsi 10, S. Corso 6, M. Crespo 3, A. Dahmani 11, V. Dangles-Marie 11, D. Decaudin 11, Z. Dudová 12, A. Fiori 6, S. Giordano 6, M. Hauptsmann 13, M. Hidalgo 14, C. Isella 6, S. de Jong 15, J. Jonkers 13, A. Křenek 12, O. Krijgsman 13, D. Kouřil 12, J.C. Lacal 14, L. Lanfrancone 16, E. Leucci 17, G.M. Mælandsmo 18, E. Marangoni 11, J. Mason 9, M.Th. Mayrhofer 19, A. Mazzocca 6, T.S. Meehan 9, E. Montaudon 11, F. Nem

Patient Opinion on the use of Artificial Intelligence in Medical Imaging
Dr Daniel H Kim