DEMOVA-package {DEMOVA}R Documentation

DEvelopment of (multi-linear QSPR/QSAR) MOdels VAlidated using test set.


Tool for the development of multi-linear QSPR/QSAR models (Quantitative structure-property/activity relationship). Theses models are used in chemistry, biology and pharmacy to find a relationship between the structure of a molecule and its property (such as activity, toxicology but also physical properties). The various functions of this package allows: selection of descriptors based of variances, intercorrelation and user expertise; selection of the best multi-linear regression in terms of correlation and robustness; methods of internal validation (Leave-One-Out, Leave-Many-Out, Y-scrambling) and external using test sets.


Package: DEMOVA
Type: Package
Version: 1.0
Date: 2016-03-15
License: GPL (>= 2)

Example of input files are avaible into the floder "tests".
# data<-read.csv("NameOfInputFile.csv",header = TRUE , sep=" ")
# mydesc<-data[,3:dim[2]]
Functions should be use in this order:
- preselection
- select_variables
- select_MLR
- fit
- LOO / LMO / Scramb (No specific order between these ones. Optional to do the rest)
- prediction
- graphe_3Sets


Vinca Prana
Maintainer: Vinca Prana <>


1. Selassie, C. D. History of Quantitative Structure-Activity Relationship; Burger's Medicinal Chemistry and Drug Discovery Sixth Edition; John Wiley & Sons Inc., 2002; Vol. 1. (2)
2. Willett, P. Chemoinformatics: a History. Wiley Interdisciplinary Reviews: Computational Molecular Science 2011, 1, 46-56.

[Package DEMOVA version 1.0 Index]