Abstract
BackgroundThe current approach to screen for drug-like molecules is to sieve for molecules with biochemical properties suitable for desirable pharmacokinetics and reduced toxicity, using predominantly biophysical properties of chemical compounds, based on empirical rules such as Lipinski's "rule of five" (Ro5). For over a decade, Ro5 has been applied to combinatorial compounds, drugs and ligands, in the search for suitable lead compounds. Unfortunately, till date, a clear distinction between drugs and non-drugs has not been achieved. The current trend is to seek out drugs which show metabolite-likeness. In identifying similar physicochemical characteristics, compounds have usually been clustered based on some characteristic, to reduce the search space presented by large molecular datasets. This paper examines the similarity of current drug molecules with human metabolites and toxins, using a range of computed molecular descriptors as well as the effect of comparison to clustered data compared to searches against complete datasets.ResultsWe have carried out statistical and substructure functional group analyses of three datasets, namely human metabolites, drugs and toxin molecules. The distributions of various molecular descriptors were investigated. Our analyses show that, although the three groups are distinct, present-day drugs are closer to toxin molecules than to metabolites. Furthermore, these distributions are quite similar for both clustered data as well as complete or unclustered datasets.ConclusionThe property space occupied by metabolites is dissimilar to that of drugs or toxin molecules, with current drugs showing greater similarity to toxins than to metabolites. Additionally, empirical rules like Ro5 can be refined to identify drugs or drug-like molecules that are clearly distinct from toxic compounds and more metabolite-like. The inclusion of human metabolites in this study provides a deeper insight into metabolite/drug/toxin-like properties and will also prove to be valuable in the prediction or optimization of small molecules as ligands for therapeutic applications.
Highlights
The current approach to screen for drug-like molecules is to sieve for molecules with biochemical properties suitable for desirable pharmacokinetics and reduced toxicity, using predominantly biophysical properties of chemical compounds, based on empirical rules such as Lipinski’s “rule of five” (Ro5)
Examining the molecular properties of three datasets The distribution of various descriptors among drugs, human metabolites and toxin molecules are available from Table 2 and Fig. 1, 2, 3, 4, based on the analyses of clustered datasets
From the analyses of clustered datasets, we find that twothirds of the human metabolites lie outside the Lipinski universe
Summary
The current approach to screen for drug-like molecules is to sieve for molecules with biochemical properties suitable for desirable pharmacokinetics and reduced toxicity, using predominantly biophysical properties of chemical compounds, based on empirical rules such as Lipinski’s “rule of five” (Ro5). This paper examines the similarity of current drug molecules with human metabolites and toxins, using a range of computed molecular descriptors as well as the effect of comparison to clustered data compared to searches against complete datasets. Druglikeness has been dominated, in the past decade, by Lipinski’s “Rule of Five” (Ro5) [3], which states that a compound is likely to be “non-drug-like” if it has more than five hydrogen bond donors, more than 10 hydrogen bond acceptors, molecular mass is greater than 500 and lipophilicity is above 5. Similar studies have spurred the quest for new approaches to classify drugs from nondrug molecules [6,7], and to characterize the properties of drug-like or lead-like compounds [8,9]. In conjunction with machine learning techniques, like Artificial Neural Networks (ANN) [6,7], Support Vector Machine (SVM) [17] and Hidden Markov Models (HMM), statistical [18] and substructure analyses have become widely accepted to characterize the properties of drug-like datasets and reduce the attrition rates in drug development
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
