Comparative Study to Predict Toxic Modes of Action Phenols

jamaicacooperativeΤεχνίτη Νοημοσύνη και Ρομποτική

17 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

96 εμφανίσεις

Comparative Study to Predict Toxic Modes of Action Phenols
from

Molecular Structures


Juan Alberto Castillo
-
Garit
a
,
b
,
d
,*

Yoan Brito
-
Sánchez
,
b,
c

Huong Le Thi Thu
,
b

Yisel
González
-
Madariaga,
c

Francisco Torrens,
d

Yovani Marrero
-
Ponce
,
b

and
J. Enrique
Rodrígu
ez
-
Borges
e


a
Centro de Estudio de Química Aplicada, Facultad de Química
-
Farmacia,
Universidad Central “Marta Abreu” de Las Villas, Santa Clara, 54830, Villa Clara,
Cuba.
juancg@uclv.edu.cu


b
Unit of Computer
-
Aided
Molecular ‘‘Biosilico’’ Discovery and Bioinformatic Research
(CAMD
-
BIR Unit), Faculty of Chemistry
-
Pharmacy,
Universidad Central “Marta
Abreu” de Las Villas
, Santa C
lara, 54830, Villa Clara, Cuba.

c
Toxicology Center, Medical Sciences University, Santa Clar
a, 50200, Villa Clara,

Cuba
.

d
Institut Universitari de Ciència Molecular, Universitat de València, Edifici d'Instituts
de Paterna, P. O. Box 22085, E
-
46071 (València), Spain.

e
Centro Investigação em Química (UP), Departamento de Química, Faculdade de
Ciênc
ias, Universidade do Porto, R. Campo Alegre, 687, P
-
4169
-
007 Porto, Portugal.



ABSTRACT


Quantitative structure

activity relationship models for the prediction of mode of
toxic action (MOA) of 221 phenols to the ciliated protozoan
Tetrahymena pyriformis
,
using atom
-
based
quadratic

indices

are reported. The phenols represent a variety of
MOAs including polar narcotics, weak acid respiratory uncouplers, pro
-
electrophiles
and soft electrophiles.

Linear discriminant analysis (LDA), and four machine learning
te
chniques (ML) [
k
-
nearest neighbors (k
-
NN), support vector machine (SVM),
classification trees (CTs) and artificial neural networks (ANNs)
] have been used to
develop several models with higher accuracies and predictive capabilities for
distinguishing betwee
n four MOAs. Most of them

showed global accuracy over 90%
and
false alarm rate
values were below 2.9% for the training set. A cross
-
validation,
complementary subsets and external test set were performed with good behavior in
all cases.

Our models compares
favorably with other previously published models
and in general, the models obtained with ML techniques shown betters results than
those developed with linear techniques. We develop unsupervised and supervised
consensus and their results were better than o
ur ML models, the results of rule
-
based approach and other ensemble model previously published. This investigation
highlights the merits of ML based techniques as an alternative to other methods more
traditional for modeling MOA.


Keywords:
Atom
-
based
Quad
ratic

Indices, Machine Learning Technique,
Quantitative Structure
-
Toxicity Relationship, Mode of Toxic Action, Phenol Derivative.