MULTIVARIATE LINEAR QSPR/QSAR MODELS:

RIGOROUS EVALUATION OF VARIABLE SELECTION FOR PLS

Computational and Structural Biotechnology Journal, 5 [6], e201302007, 1-10 (2013).

Link to this open access online journal. Direct download of PDF.

DOI: http://dx.doi.org/10.5936/csbj.201302007

Kurt Varmuza* [email, web ], Peter Filzmoser [email, web], Matthias Dehmer [email, web]

* Corresponding author

Supplementary Material: QSPR example using R

GC retention indices of polycyclic aromatic compounds (PACs),

modeled by molecular descriptors (data, software, user guide

R_for_QSPR_UserGuide.pdf User Guide (130308)

R_scripts_for_QSPR.zip R scripts (140214) New version of go_rdcv() for R 3.0.2, and errors in plots 6, 7 corrected.

PAC209_3D.zip Chemical structures of 209 PACs (polycyclic aromatic compounds) in Molfile format

contains PAC209_3D_all_H.SDF,

with approximate 3D atom coordinates and all H-atoms (by software Corina)

PAC209_dragon_2772.zip Molecular descriptors (m = 2772) calculated by software Dragon from PAC209_3D_all_H.SDF.

Contains three files:

Descriptors PAC209_dragon_2772.txt (3.5 MB)

Descriptor (variable) names PAC209_dragon_2772_descriptors.txt

Structure (object) names PAC209_dragon_2772_molecules.txt

PAC209_X2772.zip Molecular descriptors imported into R from the three Dragon output files by Dragon60_import(),

contains PAC209_X_2772.RData, load() of this RData-file gives a matrix object x (209x2772)

PAC209_X2688_y.zip Data for variable selection and PLS models (contains 2 files):

PAC209_X_2688.RData Molecular descriptors after cleaning, matrix object x (209x2688),

obtained from PAC209_X_2772.RData by varsel_almost_const()

as described in the User Guide

PAC209_y.txt Properties (GC retention indices) of the 209 PACs,

as described in the User Guide

These two files are used throughout the example for variable selection and model evaluation.

Last changes of this website 200131