Premium
Predicting Putative Inhibitors of 17β‐HSD1
Author(s) -
Heinzerling Lennart,
Hartmann Rolf W.,
Frotscher Martin,
Neumann Dirk
Publication year - 2010
Publication title -
molecular informatics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.481
H-Index - 68
eISSN - 1868-1751
pISSN - 1868-1743
DOI - 10.1002/minf.201000015
Subject(s) - chemical space , computer science , computational biology , intuition , quantitative structure–activity relationship , in silico , machine learning , data mining , drug discovery , biochemical engineering , chemistry , biochemistry , biology , engineering , philosophy , epistemology , gene
Reducing the concentration of estradiol by inhibiting its enzymatic synthesis has been proposed as a new therapeutic approach to treat estrogen‐dependent diseases. A promising potential target is 17β‐hydroxysteroid dehydrogenase type 1 which catalyzes the biosynthesis of estradiol on‐site. However, the rational computer‐aided design of novel inhibitors is still very difficult due to the scarcity of public data. Moreover, the chemical space covered in experiments has been quite limited as the design of new inhibitors was primarily guided by the intuition of experts in the field. Here, we present two different ligand‐based approaches to predict putative ligands of 17β‐hydroxysteroid dehydrogenase type 1. According to our knowledge the data set employed in our study is the largest compilation used so far allowing for thoroughly assessing the reliability of our predictive models. By combining several local models, we were able to predict putative inhibitors with an excellent expected prediction error of only 15 %. Our positive results can be considered as an encouragement for future scientific work in this field. Furthermore, the methods employed here can be easily adopted for predicting potential ligands of other enzymes.