Analysis of System Structure-Function Relationships

• Published in: ChemMedChem Vol. 2; no. 12; pp. 1774 - 1782
• Main Authors: Fliri, Anton F., Loging, William T., Volkmann, Robert A.
• Format: Journal Article
• Language: English
• Published: Weinheim WILEY-VCH Verlag 10.12.2007; WILEY‐VCH Verlag
• Subjects: biospectra; cellular response; Databases, Factual; Drug Design; proteins; Spectrum Analysis; Structure-Activity Relationship; structure-function relationships
• ISSN: 1860-7179; 1860-7187
• DOI: 10.1002/cmdc.200700153

Preclinical pharmacology studies conducted with experimental medicines currently focus on assessments of drug effects attributed to a drug's putative mechanism of action. The high failure rate of medicines in clinical trials, however, underscores that the information gathered from these studies is insufficient for forecasting drug effect profiles actually observed in patients. Improving drug effect predictions and increasing success rates of new medicines in clinical trials are some of the key challenges currently faced by the pharmaceutical industry. Addressing these challenges requires development of new methods for capturing and comparing “system‐wide” structure–effect information for medicines at the cellular and organism levels. The current investigation describes a strategy for moving in this direction by using six different descriptor sets for examining the relationship between molecular structure and broad effect information of 1064 medicines at the cellular and the organism level. To compare broad drug effect information between different medicines, information spectra for each of the 1064 medicines were created, and the similarity between information spectra was determined through hierarchical clustering. The structure–effect relationships ascertained through these comparisons indicate that information spectra similarity obtained through preclinical ligand binding experiments using a model proteome provide useful estimates for the broad drug effect profiles of these 1064 medicines in organisms. This premise is illustrated using the ligand binding profiles of selected medicines in the dataset as biomarkers for forecasting system‐wide effect observations of medicines that were not included in the incipient 1064‐medicine analysis. Improving success rates of new medicines in clinical trials requires the capture, standardization, and comparison of vast amounts of heterogeneous structure–effect information. Herein we describe an approach for relating the molecular structure of medicines to the corresponding “system‐wide” effects observed at the cellular and organism levels.