Matching Items (17)
Filtering by
- Status: Published
Description
In many classication problems data samples cannot be collected easily, example in drug trials, biological experiments and study on cancer patients. In many situations the data set size is small and there are many outliers. When classifying such data, example cancer vs normal patients the consequences of mis-classication are probably more important than any other data type, because the data point could be a cancer patient or the classication decision could help determine what gene might be over expressed and perhaps a cause of cancer. These mis-classications are typically higher in the presence of outlier data points. The aim of this thesis is to develop a maximum margin classier that is suited to address the lack of robustness of discriminant based classiers (like the Support Vector Machine (SVM)) to noise and outliers. The underlying notion is to adopt and develop a natural loss function that is more robust to outliers and more representative of the true loss function of the data. It is demonstrated experimentally that SVM's are indeed susceptible to outliers and that the new classier developed, here coined as Robust-SVM (RSVM), is superior to all studied classier on the synthetic datasets. It is superior to the SVM in both the synthetic and experimental data from biomedical studies and is competent to a classier derived on similar lines when real life data examples are considered.
ContributorsGupta, Sidharth (Author) / Kim, Seungchan (Thesis advisor) / Welfert, Bruno (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2011
Description
Reverse engineering gene regulatory networks (GRNs) is an important problem in the domain of Systems Biology. Learning GRNs is challenging due to the inherent complexity of the real regulatory networks and the heterogeneity of samples in available biomedical data. Real world biological data are commonly collected from broad surveys (profiling studies) and aggregate highly heterogeneous biological samples. Popular methods to learn GRNs simplistically assume a single universal regulatory network corresponding to available data. They neglect regulatory network adaptation due to change in underlying conditions and cellular phenotype or both. This dissertation presents a novel computational framework to learn common regulatory interactions and networks underlying the different sets of relatively homogeneous samples from real world biological data. The characteristic set of samples/conditions and corresponding regulatory interactions defines the cellular context (context). Context, in this dissertation, represents the deterministic transcriptional activity within the specific cellular regulatory mechanism. The major contributions of this framework include - modeling and learning context specific GRNs; associating enriched samples with contexts to interpret contextual interactions using biological knowledge; pruning extraneous edges from the context-specific GRN to improve the precision of the final GRNs; integrating multisource data to learn inter and intra domain interactions and increase confidence in obtained GRNs; and finally, learning combinatorial conditioning factors from the data to identify regulatory cofactors. The framework, Expattern, was applied to both real world and synthetic data. Interesting insights were obtained into mechanism of action of drugs on analysis of NCI60 drug activity and gene expression data. Application to refractory cancer data and Glioblastoma multiforme yield GRNs that were readily annotated with context-specific phenotypic information. Refractory cancer GRNs also displayed associations between distinct cancers, not observed through only clustering. Performance comparisons on multi-context synthetic data show the framework Expattern performs better than other comparable methods.
ContributorsSen, Ina (Author) / Kim, Seungchan (Thesis advisor) / Baral, Chitta (Committee member) / Bittner, Michael (Committee member) / Konjevod, Goran (Committee member) / Arizona State University (Publisher)
Created2011
Description
Given the process of tumorigenesis, biological signaling pathways have become of interest in the field of oncology. Many of the regulatory mechanisms that are altered in cancer are directly related to signal transduction and cellular communication. Thus, identifying signaling pathways that have become deregulated may provide useful information to better understanding altered regulatory mechanisms within cancer. Many methods that have been created to measure the distinct activity of signaling pathways have relied strictly upon transcription profiles. With advancements in comparative genomic hybridization techniques, copy number data has become extremely useful in providing valuable information pertaining to the genomic landscape of cancer. The purpose of this thesis is to develop a methodology that incorporates both gene expression and copy number data to identify signaling pathways that have become deregulated in cancer. The central idea is that copy number data may significantly assist in identifying signaling pathway deregulation by justifying the aberrant activity being measured in gene expression profiles. This method was then applied to four different subtypes of breast cancer resulting in the identification of signaling pathways associated with distinct functionalities for each of the breast cancer subtypes.
ContributorsTrevino, Robert (Author) / Kim, Seungchan (Thesis advisor) / Ringner, Markus (Committee member) / Liu, Huan (Committee member) / Arizona State University (Publisher)
Created2011
Description
Biological systems are complex in many dimensions as endless transportation and communication networks all function simultaneously. Our ability to intervene within both healthy and diseased systems is tied directly to our ability to understand and model core functionality. The progress in increasingly accurate and thorough high-throughput measurement technologies has provided a deluge of data from which we may attempt to infer a representation of the true genetic regulatory system. A gene regulatory network model, if accurate enough, may allow us to perform hypothesis testing in the form of computational experiments. Of great importance to modeling accuracy is the acknowledgment of biological contexts within the models -- i.e. recognizing the heterogeneous nature of the true biological system and the data it generates. This marriage of engineering, mathematics and computer science with systems biology creates a cycle of progress between computer simulation and lab experimentation, rapidly translating interventions and treatments for patients from the bench to the bedside. This dissertation will first discuss the landscape for modeling the biological system, explore the identification of targets for intervention in Boolean network models of biological interactions, and explore context specificity both in new graphical depictions of models embodying context-specific genomic regulation and in novel analysis approaches designed to reveal embedded contextual information. Overall, the dissertation will explore a spectrum of biological modeling with a goal towards therapeutic intervention, with both formal and informal notions of biological context, in such a way that will enable future work to have an even greater impact in terms of direct patient benefit on an individualized level.
ContributorsVerdicchio, Michael (Author) / Kim, Seungchan (Thesis advisor) / Baral, Chitta (Committee member) / Stolovitzky, Gustavo (Committee member) / Collofello, James (Committee member) / Arizona State University (Publisher)
Created2013
Description
Threshold logic has been studied by at least two independent group of researchers. One group of researchers studied threshold logic with the intention of building threshold logic circuits. The earliest research to this end was done in the 1960's. The major work at that time focused on studying mathematical properties of threshold logic as no efficient circuit implementations of threshold logic were available. Recently many post-CMOS (Complimentary Metal Oxide Semiconductor) technologies that implement threshold logic have been proposed along with efficient CMOS implementations. This has renewed the effort to develop efficient threshold logic design automation techniques. This work contributes to this ongoing effort. Another group studying threshold logic did so, because the building block of neural networks - the Perceptron, is identical to the threshold element implementing a threshold function. Neural networks are used for various purposes as data classifiers. This work contributes tangentially to this field by proposing new methods and techniques to study and analyze functions implemented by a Perceptron After completion of the Human Genome Project, it has become evident that most biological phenomenon is not caused by the action of single genes, but due to the complex interaction involving a system of genes. In recent times, the `systems approach' for the study of gene systems is gaining popularity. Many different theories from mathematics and computer science has been used for this purpose. Among the systems approaches, the Boolean logic gene model has emerged as the current most popular discrete gene model. This work proposes a new gene model based on threshold logic functions (which are a subset of Boolean logic functions). The biological relevance and utility of this model is argued illustrated by using it to model different in-vivo as well as in-silico gene systems.
ContributorsLinge Gowda, Tejaswi (Author) / Vrudhula, Sarma (Thesis advisor) / Shrivastava, Aviral (Committee member) / Chatha, Karamvir (Committee member) / Kim, Seungchan (Committee member) / Arizona State University (Publisher)
Created2012
Description
As we migrate into an era of personalized medicine, understanding how bio-molecules interact with one another to form cellular systems is one of the key focus areas of systems biology. Several challenges such as the dynamic nature of cellular systems, uncertainty due to environmental influences, and the heterogeneity between individual patients render this a difficult task. In the last decade, several algorithms have been proposed to elucidate cellular systems from data, resulting in numerous data-driven hypotheses. However, due to the large number of variables involved in the process, many of which are unknown or not measurable, such computational approaches often lead to a high proportion of false positives. This renders interpretation of the data-driven hypotheses extremely difficult. Consequently, a dismal proportion of these hypotheses are subject to further experimental validation, eventually limiting their potential to augment existing biological knowledge. This dissertation develops a framework of computational methods for the analysis of such data-driven hypotheses leveraging existing biological knowledge. Specifically, I show how biological knowledge can be mapped onto these hypotheses and subsequently augmented through novel hypotheses. Biological hypotheses are learnt in three levels of abstraction -- individual interactions, functional modules and relationships between pathways, corresponding to three complementary aspects of biological systems. The computational methods developed in this dissertation are applied to high throughput cancer data, resulting in novel hypotheses with potentially significant biological impact.
ContributorsRamesh, Archana (Author) / Kim, Seungchan (Thesis advisor) / Langley, Patrick W (Committee member) / Baral, Chitta (Committee member) / Kiefer, Jeffrey (Committee member) / Arizona State University (Publisher)
Created2012
Description
There are several challenges to accurately inferring levels of transcription using RNA-sequencing (RNA-seq) data, including detecting and correcting for reference genome mapping bias. One potential confounder of RNA-seq analysis results from the application of a standardized pipeline to samples of different sexes in species with chromosomal sex determination. The homology between the human X and Y chromosomes will routinely cause mismapping to occur, artificially biasing estimates of sex-biased gene transcription. For this reason we tested sex-specific mapping scenarios in humans on RNA-seq samples from the brains of 5 genetic females and 5 genetic males to assess how inferences of differential gene expression patterns change depending on the reference genome. We first applied a mapping protocol where we mapped all individuals to the entire human reference genome (complete), including the X and Y chromosomes, and computed differential expression between the set of genetic male and genetic female samples. We next mapped the genetic female samples (46,XX) to the human reference genome with the Y chromosome removed (Y-excluded) and the genetic male samples (46, XY) to the human reference genome (including the Y chromosome), but with the pseudoautosomal regions of the Y chromosome hard-masked (YPARs-masked) for the two sex-specific mappings. Using the complete and sex-specific mapping protocols, we compared the differential expression measurements of genetic males and genetic females from cuffDiff outputs. The second strategy called 33 additional genes as being differentially expressed between the two sexes when compared to the complete mapping protocol. This research provides a framework for a new standard of reference genome mappings to correct for sex-biased gene expression estimates that can be used in future studies.
ContributorsBrotman, Sarah Marie (Author) / Wilson Sayres, Melissa (Thesis director) / Crook, Sharon (Committee member) / Webster, Timothy (Committee member) / School of Life Sciences (Contributor) / School of Mathematical and Natural Sciences (Contributor) / Barrett, The Honors College (Contributor)
Created2017-05
Description
Unlike the autosomes, recombination on the sex chromosomes is limited to the pseudoautosomal regions (PARs) at each end of the chromosome. PAR1 spans approximately 2.7 Mb from the tip of the proximal arm of each sex chromosome, and a pseudoautosomal boundary between the PAR1 and non-PAR region is thought to have evolved from a Y-specific inversion that suppressed recombination across the boundary. In addition to the two PARs, there is also a human-specific X-transposed region (XTR) that was duplicated from the X to the Y chromosome. Genetic diversity is expected to be higher in recombining than nonrecombining regions, particularly because recombination reduces the effects of linked selection, allowing neutral variation to accumulate. We previously showed that diversity decreases linearly across the previously defined pseudoautosomal boundary (rather than drop suddenly at the boundary), suggesting that the pseudoautosomal boundary may not be as strict as previously thought. In this study, we analyzed data from 1271 genetic females to explore the extent to which the pseudoautosomal boundary varies among human populations (broadly, African, European, South Asian, East Asian, and the Americas). We found that, in all populations, genetic diversity was significantly higher in the PAR1 and XTR than in the non-PAR regions, and that diversity decreased linearly from the PAR1 to finally reach a non-PAR value well past the pseudoautosomal boundary in all populations. However, we also found that the location at which diversity changes from reflecting the higher PAR1 diversity to the lower nonPAR diversity varied by as much as 500 kb among populations. The lack of genetic evidence for a strict pseudoautosomal boundary and the variability in patterns of diversity across the pseudoautosomal boundary are consistent with two potential explanations: (1) the boundary itself may vary across populations, or (2) that population-specific demographic histories have shaped diversity across the pseudoautosomal boundary.
ContributorsCotter, Daniel Juetten (Author) / Wilson Sayres, Melissa (Thesis director) / Stone, Anne (Committee member) / Webster, Timothy (Committee member) / School of Life Sciences (Contributor) / School of International Letters and Cultures (Contributor) / Barrett, The Honors College (Contributor)
Created2016-12
ContributorsHesse, Eliza (Performer) / Eklund, Ansel (Performer) / Hutchins, Elizabeth (Performer) / Oakley, Ashley (Performer) / Dyches, Regina (Performer) / Xing, Wei (Performer) / Morris, Taylor (Performer) / Richardson, Jamilyn (Performer) / Freida, Tamara (Performer) / Gorbich, Victoria (Performer) / Bean, Alisha Schultz (Performer) / Tashjian, Tavit (Performer) / ASU Library. Music Library (Publisher)
Created2008-02-22
Description
Damage to the central nervous system due to spinal cord or traumatic brain injury, as well as degenerative musculoskeletal disorders such as arthritis, drastically impact the quality of life. Regeneration of complex structures is quite limited in mammals, though other vertebrates possess this ability. Lizards are the most closely related organism to humans that can regenerate de novo skeletal muscle, hyaline cartilage, spinal cord, vasculature, and skin. Progress in studying the cellular and molecular mechanisms of lizard regeneration has previously been limited by a lack of genomic resources. Building on the release of the genome of the green anole, Anolis carolinensis, we developed a second generation, robust RNA-Seq-based genome annotation, and performed the first transcriptomic analysis of tail regeneration in this species. In order to investigate gene expression in regenerating tissue, we performed whole transcriptome and microRNA transcriptome analysis of regenerating tail tip and base and associated tissues, identifying key genetic targets in the regenerative process. These studies have identified components of a genetic program for regeneration in the lizard that includes both developmental and adult repair mechanisms shared with mammals, indicating value in the translation of these findings to future regenerative therapies.
ContributorsHutchins, Elizabeth (Author) / Kusumi, Kenro (Thesis advisor) / Rawls, Jeffrey A. (Committee member) / Denardo, Dale F. (Committee member) / Huentelman, Matthew J. (Committee member) / Arizona State University (Publisher)
Created2015