Premium
PyQSAR: A Fast QSAR Modeling Platform Using Machine Learning and Jupyter Notebook
Author(s) -
Kim Sinyoung,
Cho KwangHwi
Publication year - 2019
Publication title -
bulletin of the korean chemical society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.237
H-Index - 59
ISSN - 1229-5949
DOI - 10.1002/bkcs.11638
Subject(s) - quantitative structure–activity relationship , python (programming language) , workflow , workbench , computer science , software , property (philosophy) , data mining , artificial intelligence , machine learning , database , programming language , visualization , philosophy , epistemology
Understanding the relationship between structure and property is important in current research works. The QSAR/QSPR (Quantitative Structure–Activity Relationship/Quantitative Structure–Property Relationship) is a common method for finding the relationships between the structure and property of compounds. However, traditional methods of performing QSAR analysis rely on multiple software platforms for each step. Here, an integrated standalone python package, PyQSAR, is proposed that combines all QSAR modeling process in one workbench. The efficiency of the package was verified by comparing to 10 previously published works. The results showed high performance of PyQSAR in terms of R 2 with less than half an hour execution time with a typical desktop PC for each test case. The main goal of PyQSAR is the production of reliable QSAR models on a single platform with an easy‐to‐follow workflow.