Journal: Drug Invention Today

Article Id: JPRS-Qs-00002799
Title: Quantitative structure activity relationship predictive models for antimitotic agent through machine learning on high-throughput screening of noscapine derivatives
Category: QSAR studies
Section: Research Article
  • Abstract
  • Audio Abstract
  • Authors
  • Pdf File
  • Citation
  • My Reference
  • Methodology
  • Abstract

    Background: Noscapine is a benzylisoquinoline alkaloid from plants of the poppy family, identified as an anti-mitotic agent and able to arrest mitotic cell division. Its derivatives have also proven less cytotoxic in animal models. Objective: Our objective is to identify efficient noscapine analogs which are less toxic than the existing anti-mitotic agents. Methods and Materials: We have adopted machine learning approach to screen similar structures and substructures of noscapine compounds from the PubChem database. SMILE molecular entries retrieved from the database were processed for descriptor calculation. KNIME workbench was built using modules like OpenBabel, RDKit,InchI and CanonSMILES. Descriptor calculation was done using KNIME workbench. 122 descriptors were generated for each molecule using KNIME workbench. Feature selection was done as a prior step to machine learning to eliminate inappropriate data. Two feature selection algorithms were used for attribute selection and ranking namely correlation based feature selection and information gain feature selection. Training data set containing active compounds and inactive compounds (decoy set) was prepared with ranked attributes. Mining classifiers namely Naïve Bayes, LibSVM, Random Forest and J48 were chosen to train the dataset. Cross validation was carried out for the built models. Conclusion: Among the four classifiers Random Forest has given good accuracy for our dataset. Thus an implementation of high throughput screening approach using machine learning to rank efficient noscapine derivatives has been achieved.

  • Abstract Audio

    No Audio file found

  • About the authors and Affiliations

    Author(s) Name:

    Annamalai Kaarthik, Palanisamy Balamurugan, Md. Afroz Alam*

    Affiliation(s) Name:

    Department of Biotechnology, Karunya Institute of Technology and Sciences, Coimbatore, Tamil Nadu, India

    *Corresponding author: Md. Afroz Alam, Department of Biotechnology, School of Agriculture and Biosciences, Karunya Institute of Technology and Sciences, Karunya Nagar, Coimbatore - 641 114, Tamil Nadu, India

  • View Article File in pdf format.

    Article File
  • View Article Citation Here.

    0 View More
  • How to Cite my Article.

    Author:

    Annamalai Kaarthik, Palanisamy Balamurugan, Md. Afroz Alam*

    Title:Quantitative structure activity relationship predictive models for antimitotic agent through machine learning on high-throughput screening of noscapine derivatives
    Journal:Drug Invention Today
    Vol(issue):10 (October [ Special Issue 2 ])
    Year:2018
    Page No: (3003-3008)
  • Experimental Methods Keywords

    Methodology:Quantitative structure activity relationship
    Research Materials:Noscapine

Keywords

Classifiers Descriptors Machine learning Microtubules Noscapine Quantitative structure activity relationship

Our Services

Most Downloaded List