Classification of dopamine, serotonin, and dual antagonists by decision trees

Hye Jung Kim, Hyunah Choo, Yong Seo Cho, Hun Yeong Koh, Kyoung Tai No, Ae Nim Pae

Research output: Contribution to journalArticle

15 Citations (Scopus)

Abstract

Dopamine antagonists (DA), serotonin antagonists (SA), and serotonin-dopamine dual antagonists (Dual) are being used as antipsychotics. A lot of dopamine and serotonin antagonists reveal non-selective binding affinity against these two receptors because the antagonists share structurally common features originated from conserved residues of binding site of the aminergic receptor family. Therefore, classification of dopamine and serotonin antagonists into their own receptors can be useful in the designing of selective antagonist for individual therapy of antipsychotic disorders. Data set containing 1135 dopamine antagonists (D2, D3, and D4), 1251 serotonin antagonists (5-HT1A, 5-HT2A, and 5-HT 2C), and 386 serotonin-dopamine dual antagonists was collected from the MDDR database. Cerius2 descriptors were employed to develop a classification model for the 2772 compounds with antipsychotic activity. LDA (linear discriminant analysis), SIMCA (soft independent modeling of class analogy), RP (recursive partitioning), and ANN (artificial neural network) algorithms successfully classified the active class of each compound at the average 73.6% and predicted at the average 69.8%. The decision trees from RP, the best model, were generated to identify and interpret those descriptors that discriminate the active classes more easily. These classification models could be used as a virtual screening tool to predict the active class of new candidates.

Original languageEnglish
Pages (from-to)2763-2770
Number of pages8
JournalBioorganic and Medicinal Chemistry
Volume14
Issue number8
DOIs
Publication statusPublished - 2006 Apr 15

Fingerprint

Decision Trees
Serotonin Antagonists
Dopamine Antagonists
Decision trees
Dopamine
Serotonin
Antipsychotic Agents
Discriminant analysis
Discriminant Analysis
Screening
Binding Sites
Neural networks
Databases

All Science Journal Classification (ASJC) codes

  • Biochemistry
  • Molecular Medicine
  • Molecular Biology
  • Pharmaceutical Science
  • Drug Discovery
  • Clinical Biochemistry
  • Organic Chemistry

Cite this

Kim, Hye Jung ; Choo, Hyunah ; Cho, Yong Seo ; Koh, Hun Yeong ; No, Kyoung Tai ; Pae, Ae Nim. / Classification of dopamine, serotonin, and dual antagonists by decision trees. In: Bioorganic and Medicinal Chemistry. 2006 ; Vol. 14, No. 8. pp. 2763-2770.
@article{0a3615d0987749a2bbf71faf423dc516,
title = "Classification of dopamine, serotonin, and dual antagonists by decision trees",
abstract = "Dopamine antagonists (DA), serotonin antagonists (SA), and serotonin-dopamine dual antagonists (Dual) are being used as antipsychotics. A lot of dopamine and serotonin antagonists reveal non-selective binding affinity against these two receptors because the antagonists share structurally common features originated from conserved residues of binding site of the aminergic receptor family. Therefore, classification of dopamine and serotonin antagonists into their own receptors can be useful in the designing of selective antagonist for individual therapy of antipsychotic disorders. Data set containing 1135 dopamine antagonists (D2, D3, and D4), 1251 serotonin antagonists (5-HT1A, 5-HT2A, and 5-HT 2C), and 386 serotonin-dopamine dual antagonists was collected from the MDDR database. Cerius2 descriptors were employed to develop a classification model for the 2772 compounds with antipsychotic activity. LDA (linear discriminant analysis), SIMCA (soft independent modeling of class analogy), RP (recursive partitioning), and ANN (artificial neural network) algorithms successfully classified the active class of each compound at the average 73.6{\%} and predicted at the average 69.8{\%}. The decision trees from RP, the best model, were generated to identify and interpret those descriptors that discriminate the active classes more easily. These classification models could be used as a virtual screening tool to predict the active class of new candidates.",
author = "Kim, {Hye Jung} and Hyunah Choo and Cho, {Yong Seo} and Koh, {Hun Yeong} and No, {Kyoung Tai} and Pae, {Ae Nim}",
year = "2006",
month = "4",
day = "15",
doi = "10.1016/j.bmc.2005.11.059",
language = "English",
volume = "14",
pages = "2763--2770",
journal = "Bioorganic and Medicinal Chemistry",
issn = "0968-0896",
publisher = "Elsevier Limited",
number = "8",

}

Classification of dopamine, serotonin, and dual antagonists by decision trees. / Kim, Hye Jung; Choo, Hyunah; Cho, Yong Seo; Koh, Hun Yeong; No, Kyoung Tai; Pae, Ae Nim.

In: Bioorganic and Medicinal Chemistry, Vol. 14, No. 8, 15.04.2006, p. 2763-2770.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Classification of dopamine, serotonin, and dual antagonists by decision trees

AU - Kim, Hye Jung

AU - Choo, Hyunah

AU - Cho, Yong Seo

AU - Koh, Hun Yeong

AU - No, Kyoung Tai

AU - Pae, Ae Nim

PY - 2006/4/15

Y1 - 2006/4/15

N2 - Dopamine antagonists (DA), serotonin antagonists (SA), and serotonin-dopamine dual antagonists (Dual) are being used as antipsychotics. A lot of dopamine and serotonin antagonists reveal non-selective binding affinity against these two receptors because the antagonists share structurally common features originated from conserved residues of binding site of the aminergic receptor family. Therefore, classification of dopamine and serotonin antagonists into their own receptors can be useful in the designing of selective antagonist for individual therapy of antipsychotic disorders. Data set containing 1135 dopamine antagonists (D2, D3, and D4), 1251 serotonin antagonists (5-HT1A, 5-HT2A, and 5-HT 2C), and 386 serotonin-dopamine dual antagonists was collected from the MDDR database. Cerius2 descriptors were employed to develop a classification model for the 2772 compounds with antipsychotic activity. LDA (linear discriminant analysis), SIMCA (soft independent modeling of class analogy), RP (recursive partitioning), and ANN (artificial neural network) algorithms successfully classified the active class of each compound at the average 73.6% and predicted at the average 69.8%. The decision trees from RP, the best model, were generated to identify and interpret those descriptors that discriminate the active classes more easily. These classification models could be used as a virtual screening tool to predict the active class of new candidates.

AB - Dopamine antagonists (DA), serotonin antagonists (SA), and serotonin-dopamine dual antagonists (Dual) are being used as antipsychotics. A lot of dopamine and serotonin antagonists reveal non-selective binding affinity against these two receptors because the antagonists share structurally common features originated from conserved residues of binding site of the aminergic receptor family. Therefore, classification of dopamine and serotonin antagonists into their own receptors can be useful in the designing of selective antagonist for individual therapy of antipsychotic disorders. Data set containing 1135 dopamine antagonists (D2, D3, and D4), 1251 serotonin antagonists (5-HT1A, 5-HT2A, and 5-HT 2C), and 386 serotonin-dopamine dual antagonists was collected from the MDDR database. Cerius2 descriptors were employed to develop a classification model for the 2772 compounds with antipsychotic activity. LDA (linear discriminant analysis), SIMCA (soft independent modeling of class analogy), RP (recursive partitioning), and ANN (artificial neural network) algorithms successfully classified the active class of each compound at the average 73.6% and predicted at the average 69.8%. The decision trees from RP, the best model, were generated to identify and interpret those descriptors that discriminate the active classes more easily. These classification models could be used as a virtual screening tool to predict the active class of new candidates.

UR - http://www.scopus.com/inward/record.url?scp=33644784708&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33644784708&partnerID=8YFLogxK

U2 - 10.1016/j.bmc.2005.11.059

DO - 10.1016/j.bmc.2005.11.059

M3 - Article

C2 - 16387502

AN - SCOPUS:33644784708

VL - 14

SP - 2763

EP - 2770

JO - Bioorganic and Medicinal Chemistry

JF - Bioorganic and Medicinal Chemistry

SN - 0968-0896

IS - 8

ER -