Robert H. Herring

Multivariate characterization, modeling, and design of ionic liquid molecules

Abstract Ionic liquids that have tailored structures with an array of unique functional properties can have important applications in areas such as CO2 capture and sequestration, sulfur removal from fuels, energy storage, biomass pretreatment, and chemical separations. Within a computer-aided molecular design (CAMD) framework, a characterization based method was combined with chemometric techniques in a reverse problem formulation to design ionic liquid (IL) structures corresponding to particular physical properties. Infrared spectra generated from density functional theory (DFT) simulations were used for capturing information on molecular architecture and calibration of latent variable property models to synthesize ILs in a logical and systematic methodology.

Design of Ionic Liquids Using Property Clustering and Decomposition Techniques

Abstract Ionic liquids that have tailored structures with an array of unique functional properties can have important applications in areas such as C0 2 capture and sequestration, sulfur removal from fuels, energy storage, biomass pretreatment, and chemical separations. Within the computer-aided molecular design (CAMD) framework, a characterization based method was combined with chemometric and property clustering techniques in reverse problem formulation to the design of ionic liquid (IL) structures corresponding to particular physical properties. Infrared spectra generated from density functional theory simulation were used for capturing information on molecular architecture and calibration of latent variable property models to synthesize ILs in a logical and systematic methodology.

Evolutionary algorithm for de novo molecular design with multi-dimensional constraints

Abstract An evolutionary approach for solving molecular design problems with descriptors of varying dimensionality has been developed. Spatial fragment based descriptors are employed to generate candidate solutions within a population, which is evolved through the application of genetic operators toward an improved fitness. The candidate molecules are represented as graphs, and as such, customized operators of crossover and mutation have been developed to be compatible with this representation. The search space is conveniently represented through limitations on the occurrence of each fragment, as defined by the chosen data set, and the spatial capabilities of this space are captured through an initial conformational analysis. This spatial information is compressed and utilized to generate conformational space estimations throughout the algorithm, which expedites the search for solution graphs. The effect of various user determined input parameters is considered and exemplified through a case study involving the identification of solvents falling within a desired boiling point range, as estimated by a multi-dimensional property model.

De Novo Molecular Design using a Graph-Based Genetic Algorithm Approach

Abstract The area of computer-aided molecular design has greatly influenced the rate and cost at which novel chemicals with desired attributes have XDbeen identified. As such, great effort has been invested in new methodologies which allow for the solution of larger and more complex problems of this nature. The application of genetic algorithms (GAs) is one such technique which has shown promise in the solution of large combinatorial, and highly non-linear molecular design problems. In addition, it has been shown that many molecular properties or attributes are often best characterized by a combination of descriptors with varying dimensionality. The inverse solution to property models of this nature, which entails identifying candidate molecular structures with the desired properties as defined by the given model, is often highly non-linear in nature. In addition, the use of molecular fragments, as often practiced in the de novo design of novel structures, can lead to a combinatorially large search space which becomes intractable for exhaustive solution techniques. The application of GAs provides a powerful method for the solution of these types of molecular design problems in which there are often multiple objective constraints with high computational complexity and a large search space. This approach utilizes a fragment based descriptor known as the signature descriptor, which is represented as a molecular graph, as building blocks to generate candidate solutions. The graph-based genetic operators necessary for such an approach will be outlined as well as exemplified through a case study which will highlight the advantages of this algorithm.

Data Mining and Regression Algorithms for the Development of a QSPR Model Relating Solvent Structure and Ibuprofen Crystal Morphology

Abstract Computer Aided Molecular Design (CAMD) has been used to supplement and guide experimental efforts in various fields. Currently, highly predictive models can be developed as a result of improved computing hardware along with novel methodologies for model development. Decreasing dependency on experimental efforts can also reduce the environmental footprint of companies involved in product design and development. Though being a well-practiced process, crystallization is comparatively a much less studied separation unit operation. Not much is known about the variables that affect quality and usefulness of the end product gained from crystallization. Studies suggest that the interactions between solute and solvent are quite difficult to quantify and large variance can be noticed with every solute-solvent combination. Pharmaceutical companies are widely known for using crystallization operations to develop their end product. Additionally, crystal morphology has a significant impact on how the drug is metabolized in the human body. This contribution describes the development of a quantitative structure-property relationship (QSPR) model which relates the crystal aspect ratio of ibuprofen to the structure of the solvent utilized. The work focuses on applying various data mining and regression algorithms such as PCA/PCR (Principal Component Analysis/Regression), GA-MLR (Genetic Algorithm - Multiple Linear Regression), ANN (Artificial Neural Network) to generate a QSPR model. The validation results of the models generated from these algorithms are thoroughly analyzed and explained. Three-dimensional (3D) descriptors are often found to cause degeneracy issues when applied within QSPR studies. For these reasons, singular value decomposition (SVD) was used as an alternative to PCA for its increased stability in primary component prediction. Both internal and external validation methods were used to test the predictive capabilities of the developed models.

Molecular Design using Three-Dimensional Signature Descriptors

Abstract Integrated process and product design is a useful approach for identifying globally optimal solutions to increasingly demanding problems. Property based process and product design methods are convenient since both sides of the problem can easily be expressed in terms of molecular properties. Property clustering, a technique for tracking stream properties or functionalities through conserved quantities, allows for interpretation of process performance as a function of properties. The molecular signature descriptor is quite effective in this approach as any topological index, which enumerates molecular information for use in QSPRs, can be derived from the molecular signature. Previous works on application of the signature descriptor have been limited to use with topological indices. This contribution outlines an algorithm for including topographical, or three-dimensional, information with the signature descriptor and how to apply this information in the reverse problem formulation (RPF) methodology.

Incorporating Topographical Characteristics in Molecular Signature Descriptors

Abstract Several techniques have been developed in recent years for integrated process and product design. The recently introduced concept known as molecular signature descriptors has been used in our previous works to design molecules that meet the desired property targets. This contribution extends the applicability of the signature descriptor to include 3D molecular information and outlines the algorithm necessary for applying it in the reverse problem formulation method (RPF). The technique introduces the three dimensional character of a graph in the form of a molecular geometry matrix, which can be manipulated to provide several useful descriptors. The ability to include topological, as well as topographical, information when developing a QSAR on the molecular signature platform provides a powerful tool for solving problems using the reverse formulation approach.

Graph-Based Genetic Algorithm for De Novo Molecular Design

Abstract The field of computer-aided molecular design has benefited from the introduction of many new methodologies made possible through improvements in computing capabilities and techniques for handling large amounts of data. These advances in computational capabilities have ushered in an array of improved molecular simulation techniques as well as ways to characterize these phenomena. The increase in useful descriptors, capturing more detailed information than ever before, has allowed for more flexibility in developing structure-property and activity relationships. Coupled with powerful variable selection techniques, increasingly accurate and predictive models are being generated. This requires a paralleled increase in methodologies useful for utilization of these models, often containing descriptors of widely varying dimensionality, in a predictive, or inverse, manner. The application of genetic algorithms (GAs) is one such technique which has shown promise in the solution of large combinatorial, and highly non-linear molecular design problems. The approach presented here utilizes a fragment based descriptor known as the Signature descriptor, which is represented as a molecular graph, as building blocks to generate candidate solutions. The graph-based genetic operators necessary for such an approach will be outlined as well as exemplified through a case study which will highlight the advantages of the algorithm.

Characterization Based Reverse Design of Ionic Liquids

Ionic liquids (ILs) that have tailored structures with an array of unique functional properties can have important applications in areas such as CO2 capture and sequestration, sulfur removal from fuels, energy storage, biomass pretreatment, and chemical separations. Within a computer-aided molecular design (CAMD) framework, a characterization based method is combined with chemometric and property clustering techniques in a reverse problem formulation for the design of ionic liquid (IL) structures with specific physical properties. Infrared spectra generated from density functional theory simulation were used for capturing information on molecular architecture and calibration of latent variable property models to synthesize ILs.

Development of QSPR Model Relating Solvent Structure to Crystal Morphology

Abstract This contribution outlines the development of a quantitative structure-property relationship (QSPR) that relates solvent structure to the morphology of ibuprofen crystals grown within that solvent. Morphology can be quantified by aspect ratio, and ibuprofen aspect ratio data was obtained for crystals grown in 16 different organic solvents. A combination of 2D and 3D molecular descriptors are calculated to provide a quantitative representation of the geometry optimized solvent molecules. Empirical force fields were used to estimate the three-dimensional structure of the solvent molecules and three different force fields are implemented while their effect on the developed models is analyzed. The descriptor data matrix, containing a multitude of descriptor types, is reduced in size, using Bayesian Information Criterion (BIC) methods and also Principal Component Analysis (PCA), for regression into linear models through Principal Component Regression. The predictive capabilities of these models were also analyzed through means of internal and external validation methods.