Evolutionary synthesis of stochastic gene network models using feature-based search spaces

A feature-based fitness function is applied in a genetic programming system to synthesize stochastic gene regulatory network models whose behaviour is defined by a time course of protein expression levels. Typically, when targeting time series data, the fitness function is based on a sum-of-errors involving the values of the fluctuating signal. While this approach is successful in many instances, its performance can deteriorate in the presence of noise. This thesis explores a fitness measure determined from a set of statistical features characterizing the time series' sequence of values, rather than the actual values themselves. Through a series of experiments involving symbolic regression with added noise and gene regulatory network models based on the stochastic 'if-calculus, it is shown to successfully target oscillating and non-oscillating signals. This practical and versatile fitness function offers an alternate approach, worthy of consideration for use in algorithms that evaluate noisy or stochastic behaviour.

Spectral Reflectance Characteristics of Kimberlite Indicator Minerals and Other Selected Mineral Grains

High chromium content in kimberlite indicator minerals such as pyrope garnet and diopside is often correlated with the presence of diamonds. In this study, kimberlite indicator minerals were examined using visible light reflectance spectroscopy to determine if chromium content can be correlated with spectral absorption features. The depth of absorption features in the visible spectral region were correlated with the molecular percentage of chromium and other first series transition metal elements obtained by electron microprobe data. In the visible part of the spectrum, chromium is evident by 3 absorption features in the pyrope reflectance spectrum; one isolated and narrow feature at the wavelength 689 nm was used to correlate with the chromium mol %. The isolation of this feature in the pyrope spectra is advantageous since it is not directly affected by other proximal absorption bands that could be caused by other transition metals. Analysis of the feature indicates that as grain volume increases the depth of the absorption feature will also increase. Clustering grain volumes into fractions yields better correlation between absorption depth and mol % chromium. Other types of garnet (almandine, grossular, spessartine) and kimberlite indicator minerals (olivine, diopside, chromite, ilmenite) were analyzed to determine if other absorption features could be used to predict the proportion of specific transition metal elements. Diopside in particular illustrates the same isolated chromium absorption feature as pyrope and may indicate mol percent but needs further study with larger sample sets.

Two Levels of Motivation and Two Types of Well-Being: Relations Between General and Goal-Specific Motivations, and Eudaimonic and Hedonic Well-Being

In this thesis I assess the individual and joint predictive associations and effects between multiple motivation and well-being concepts. In particular, three pairs of motivation concepts (intrinsic/extrinsic, approach/avoidance, and eudaimonic/hedonic) are assessed simultaneously at two levels of analysis (disposition and goal) and examined in relation to two types of well-being (eudaimonic and hedonic) in two studies, one correlational and the other experimental. Study 1: Using a correlational design, participants (N = 325, M age = 19.10, 87% female) completed self-report measures assessing six motivation and two well-being concepts. Exploratory factor analyses were used to assess patterns of associations among the motivational constructs. Results indicated that constructs displaying conceptual and empirical similarities co-occur, particularly, intrinsic, approach and eudaimonic motivation. Regression models were used to assess predictive relations between the motivational constructs and well-being. Both types of well-being were predicted by approach and avoidance dispositions, and hedonic goals. Additionally, eudaimonic well-being was uniquely predicted by eudaimonic dispositions and goals, and intrinsic dispositions; and hedonic well-being was uniquely predicted by extrinsic dispositions and approach goals. The patterns of associations among motivational constructs, and similarities and differences in the ways they predict each type of well-being, are discussed. Study 2: Using an experimental design, participants (N = 447, M age = 19.30, 88% female) were randomly assigned to one of eight experimental conditions, each involving a manipulation aimed at priming combinations of the three pairs of motivational constructs at the goal level. Participants then completed measures of both types of well-being. ANOVAs were used to assess the main effects and interactions of experimental condition for each of the three pairs of motivational constructs on well-being. Main effects of experimental conditions were non-significant. However, results indicated that focus on each of the three pairs of motivational constructs predicted well-being and that the manipulation impacted well-being indirectly, through experimentally-shifted motivational focus. Few interactions emerged. Implications for future experimental research and the conceptual integration of motivation and well-being constructs are discussed. In conclusion, Studies 1 and 2 inform the motivation and well-being fields in novel ways and provide preliminary steps towards studying these fields from an integrated and comprehensive motivational framework.

UNDERSTANDING THE RELATIONSHIP BETWEEN ONCOLYTIC AD5 DELETED E1b55KDA LYTIC INFECTION AND P53 IN MAMMALIAN CELLS

Adenoviruses are the most commonly used in the development of oncolytic therapy. Oncolytic adenoviruses are genetically modified to selectivity replicate in and kill tumor cells. The p53 molecule is a tumor suppressor protein that responds to viral infection through the activation of apoptosis, which is inhibited by adenovirus E1B55kDa protein leading to progressive viral lytic cycle. The non-specificity of replication has limited the use of wild type adenovirus in cancer therapy. This issue was resolved by using an E1b deleted Ad that can only replicate in cells with a deficiency in the p53 protein, a common feature of most cancer cells. Although demonstrating a moderate success rate, E1b55kDa deleted Ad has not been approved as a standard therapy for all cancer types. Several studies have revealed that E1b deleted Ad replication was independent of p53 status in the cell, as the virus replicated better in some p53 deficient cancers more than others. However, this mechanism has not been investigated deeply. Therefore, the objective of this study is to understand the relationship between p53 status, levels and functional activity, and oncolytic Ad5dlE1b55kDa replication efficiency. Firstly, five transient p53 expression vectors that contain different regulatory elements were engineered and then evaluated in H1299, HEK293 and HeLa cell lines. Data indicated that vector that contains the MARs and HPRE regulatory elements achieved the highest stability of p53 expression. Secondly, we used these vectors to examine the effect of various p53 expression levels on the replication efficiency of oncolytic Ad5dlE1b55kDa. We found that the level of p53 in the cell had an insignificant effect on the oncolytic viruses’ replication. However, the functional activity of p53 had a significant effect on its replication, as Ad5dlE1b55kDa was shown to have selective activity in H1299 cells (p53-null). In contrast, a decrease in viral replication was found in HeLa cells (p53-positive). Finally, the effect of p53’s functional activity on the replication efficiency of oncolytic Ad5dlE1b55kDa was examined. Viral growth was evaluated in H1299 cells expressing number of p53 mutants. P53-R175H mutant successfully rescued viral growth by allowing the virus to exert its mechanism of selectivity. The mechanism entailed deregulating the expression of specific genes, cell cycle and apoptosis, in the p53 pathway to promote its production leading to efficient oncolytic effect. These results confirmed that oncolytic Ad5dlE1b55kDa sensitivity is mutation-type specific. Therefore, before it is applied clinically as cancer therapy for p53 deficient tumors, the type of p53 mutation must be determined for efficient antitumor effect.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and determinis- tic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel meta–heuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS meta–heuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

New Feature at Niagara – Clark Hill Islands

New Feature at Niagara – Clark Hill Islands (5 islands situated in the rapids of the Niagara River). These islands are currently known as Dufferin Islands, 22 ½ cm. x 15 ½ cm, n.d.