Molecular similarity/toxicity model of benzene derivatives and a similarity analysis of environmentally impacted regions.
Abstract
A fundamental principle of chemistry and biochemistry is that chemical properties and biological activities are related, if not dictated, by chemical structure at the molecular and polymolecular levels. An important tool in understanding the nature of property/activity dependence on molecular structure is a research protocol known by the acronym QSAR. In Quantitative Structure Activity Relationships, the numerical values of experimental properties or empirically derived parameters are used as independent variables to correlate observed values of biological activities.
The fundamental postulate of QSAR analysis is that compounds with similar molecular structures will exhibit similar chemical/biological properties. It is necessary to have a reliable, practical, and rational means by which to assess similarity---thus far this has been impossible. In the first part of this dissertation molecular QSAR descriptors were used to derive numerical measures of molecular similarity which allowed the postulate to be rigorously tested. This was accomplished by using similarity measures to rectify the experimental values of a toxicological property for a data set of 485 benzene derivatives. This quantitative method for assessing similarity can be carried out using readily available statistical programs, and proved capable of correlating 60.8% of the variance in the data, compared to 58.8% for the best QSAR model. The development of this similarity method, which is termed Quantitative Structure Similarity Analysis (QSSA), provides a new tool to use in the development of new drugs, or to employ in structure/activity/mechanism studies. In addition, an examination of presently utilized validation techniques was carried out, leading to the discovery that permutation testing is the most stringent assessment of a model's correlative ability.
The second part of this dissertation extended the concepts of QSAR/similarity analyses to environmental science. A new application for similarity was demonstrated with the same types of statistical techniques applied to the QSAR study. To illustrate the overall approach, demographic data for 9 U.S. cities were collected. Similarity matrices created from the data provided a numerical value for the similarity relationship between pairs of cities. Applying the findings from such comparisons, environmental assessments could be made in order to create a more suitable means by which environmental compliance could be assessed for any air quality, water quality, or nuclear/industrial/chemical waste issue.
Subject Area
Environmental Sciences; Chemistry, Organic; Chemistry, Analytical
Recommended Citation
Rum, Gabrielle, "Molecular similarity/toxicity model of benzene derivatives and a similarity analysis of environmentally impacted regions." (1999). ETD Collection for University of Texas, El Paso. AAI9959917.
https://scholarworks.utep.edu/dissertations/AAI9959917