Premium
Focused hierarchical design of peptide libraries—follow the lead
Author(s) -
Muthas Daniel,
Lek Per M.,
Nurbo Johanna,
Karlén Anders,
Lundstedt Torbjörn
Publication year - 2007
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.1069
Subject(s) - dipeptide , quantitative structure–activity relationship , computer science , peptide , artificial intelligence , combinatorial chemistry , chemistry , computational biology , machine learning , biology , biochemistry
A novel design strategy based on the hierarchical design of experiments (HDoE) method named focused hierarchical design of experiments (FHDoE) is presented. FHDoE combine two design layers and use focused substitutions to increase the probability of obtaining active peptides when designing libraries through a selection of compounds biased towards a lead structure. Increasing the number of peptides with measurable activity will increase the information gained and the likelihood of constructing good quantitative structure–activity relationship (QSAR) models. The utility of the novel design method is verified using two different approaches. First, a library designed with the novel FHDoE method was compared with libraries generated from classical positional scanning techniques (e.g., alanine scan) as well as with general and centered minimum analog peptide sets (MAPS) libraries by using an example found in the literature. Secondly, the same design strategies were applied to a dataset of 58 angiotensin converting enzyme (ACE) dipeptide inhibitors. QSAR models were generated from designed sublibraries and the activities of the remaining compounds were predicted. These two examples show that the use of FHDoE renders peptide libraries close in physicochemical space to the native ligand, yielding a more thorough screening of the area of interest as compared to the classical positional scans and fractional factorial design (FFD). It is also shown that an FHDoE library of six dipeptides could produce a QSAR model that better described the requisites of high activity ACE inhibitors than could QSAR models built from either a nine‐dipeptide library designed with MAPS or a 58‐dipeptide library. Copyright © 2007 John Wiley & Sons, Ltd.