Browsing by Subject "Transformation"

Modeling of extreme data often involves thresholding, or retaining only the most extreme observations, in order that the tail may "speak" and not be overwhelmed by the bulk of the data. We describe a transformation-based framework that allows univariate density estimation to smoothly transition from a flexible, semi-parametric estimation of the bulk into a parametric estimation of the tail without thresholding. In the limit, this framework has desirable theoretical tail-matching properties to the selected parametric distribution. We develop three Bayesian models under the framework: one using a logistic Gaussian process (LGP) approach; one using a Dirichlet process mixture model (DPMM); and one using a predictive recursion approximation of the DPMM. Models produce estimates and intervals for density, distribution, and quantile functions across the full data range and for the tail index (inverse-power-decay parameter), under an assumption of heavy tails. For each approach, we carry out a simulation study to explore the model's practical usage in non-asymptotic settings, comparing its performance to methods that involve thresholding.

Among the three models proposed, the LGP has lowest bias through the bulk and highest quantile interval coverage generally. Compared to thresholding methods, its tail predictions have lower root mean squared error (RMSE) in all scenarios but the most complicated, e.g. a sharp bulk-to-tail transition. The LGP's consistent underestimation of the tail index does not hinder tail estimation in pre-extrapolation to moderate-extrapolation regions but does affect extreme extrapolations.

An interplay between the parametric transform and the natural sparsity of the DPMM sometimes causes the DPMM to favor estimation of the bulk over estimation of the tail. This can be overcome by increasing prior precision on less sparse (flatter) base-measure density shapes. A finite mixture model (FMM), substituted for the DPMM in simulation, proves effective at reducing tail RMSE over thresholding methods in some, but not all, scenarios and quantile levels.

The predictive recursion marginal posterior (PRMP) model is fast and does the best job among proposed models of estimating the tail-index parameter. This allows it to reduce RMSE in extrapolation over thresholding methods in most scenarios considered. However, bias from the predictive recursion contaminates the tail, casting doubt on the PRMP's predictions in tail regions where data should still inform estimation. We recommend the PRMP model as a quick tool for visualizing the marginal posterior over transformation parameters, which can aid in diagnosing multimodality and informing the precision needed to overcome sparsity in the mixture model approach.

In summary, there is not enough information in the likelihood alone to prevent the bulk from overwhelming the tail. However, a model that harnesses the likelihood with a carefully specified prior can allow both the bulk and tail to speak without an explicit separation of the two. Moreover, retaining all of the data under this framework reduces quantile variability, improving prediction in the tails compared to methods that threshold.

Reversible acetylation has emerged as an important post-translational modification that rivals phosphorylation in regulating chromatin structure and gene expression. Acetylation of histone is associated with transcriptional activation while deacetylation is linked to transcriptional repression. Moreover, histone deacetylase inhibitors induce growth arrest, differentiation and apoptosis of cancer cells and therefore appear to be promising anti-tumor agents. While transcriptional deregulation is thought to be the main mechanism underlying their therapeutic effects, the exact mechanisms and targets by which HDAC inhibitors, which are mostly non-specific, achieve their anti-tumor effects remain poorly understood. In other words, it is not known which and how each HDAC members are involved in supporting tumor growth.

In this thesis, I have showed that HDAC6, a cytoplasmic localized and cytoskeleton-associated deacetylase, is required for efficient oncogenic transformation and tumor formation. I have found that HDAC6 expression is induced upon oncogenic Ras-induced transformation in both human somatic cells and murine fibroblasts. Conversely, murine fibroblasts deficient in HDAC6 are more resistant to both oncogenic Ras and ErbB2-dependent transformation, indicating a critical role for HDAC6 in oncogene-induced transformation. Supporting this hypothesis, inactivation of HDAC6 in several human cancer cell lines effectively impairs anchorage-independent growth in vitro and their ability to form tumors in immunocompromised mice. I have demonstrated that the impairment of anchorage independent growth in HDAC6 deficient cells is associated with increased anoikis and mechanistically a defect in activation of the AKT and ERK kinase cascades. Additionally, HDAC6 null mice are more resistant to two-stage chemical carcinogen-induced skin tumors. Finally, I have demonstrated that the tumor-promoting effect of HDAC6 is probably mediated through the molecular chaperon Hsp90. While Hsp90 is known to be deacetylated by HDAC6 and has been implicated in stabilizing Raf-1 and ErbB2, I have found that suppression of HDAC6 impairs the stability of Raf-1 and the association between Hsp90 and ErbB2.

In conclusion, my work provides the first experimental evidence that of all the HDAC members, the cytoplasmic deacetylase HDAC6 is required for efficient oncogenic transformation, indicating that reversible acetylation plays a critical role in regulating cellular functions of non-histone non-nuclear cytoplasmic proteins that contributes to malignant transformation. Furthermore, this work identifies HDAC6 as an important component underlying the anti-tumor effects of HDAC inhibitors.

Electrowetting-on-dielectric (EWD) digital microfluidics is a droplet-based fluid handling technology capable of radically accelerating the pace of genome engineering research. EWD-based laboratory-on-chip (LoC) platforms demonstrate excellent performance in automating labor-intensive laboratory protocols at ever smaller scales. Until now, there has not been an effective means of gene transfer demonstrated in EWD microfluidic platforms. This thesis describes the theoretical and experimental approaches developed in the demonstration of an EWD-enabled electrotransfer device. Standard microfabrication methods were employed in the integration of electroporation (EP) and EWD device architectures. These devices enabled the droplet-based bulk transformation of E. coli with plasmid and oligo DNA. Peak on-chip transformation efficiencies for the EP/EWD device rivaled that of comparable benchtop protocols. Additionally, ultrasound induced in-droplet microstreaming was developed as a means of improving on-chip electroporation. The advent of electroporation in an EWD platform offers synthetic biologists a reconfigurable, programmable, and scalable fluid handling platform capable of automating next-generation genome engineering methods. This capability will drive the discovery and production of exotic biomaterials by providing the instrumentation necessary for rapidly generating ultra-rich genomic diversity at arbitrary volumetric scales.

The genes encoding the Ras family of small GTPases are mutated to yield constitutively active GTP-bound oncoproteins in one-third of all human cancers. In many other cancers lacking Ras mutations, Ras is activated by other means. One common example of such activation is found in breast cancer, in which epidermal growth factor receptor (EGFR) family receptor tyrosine kinases, including EGFR and HER2 (ErbB-2/Neu), are frequently amplified and overexpressed, which in turn activates Ras. In human cells, activation of the Ral guanine nucleotide exchange factor, or RalGEF, effector pathway is necessary for Ras-mediated tumorigenesis and metastasis. RalGEFs activate the two highly similar Ral GTPases, RalA and RalB. While RalA has been shown to be required for Ras-mediated tumorigenesis, RalB is important for tumor metastasis. Activated Ral GTPases bind to and activate a limited number of effector proteins, including RalBP1, Sec5, and Exo84, to affect numerous diverse activities of the cell. This dissertation research sought to determine which of these well-characterized Ral effector proteins were required for oncogenic mutant Ras-induceded tumorigenesis and metastasis of human cells, as well as to examine the role of RalA in breast cancer cells that can activate Ras through EGFR and HER2 overexpression.

RNA interference-mediated loss-of-function analysis demonstrated that Sec5 and Exo84 are required for oncogenic Ras-mediated tumorigenesis, and, at least in part, metastasis. Additionally, both gain-of-function and inhibition studies showed that RalA activation is induced by EGFR and HER2 in breast cancer cell lines stimulated with EGF. Furthermore, stable suppression of RalA expression inhibited tumorigenic growth of breast cancer cells, and RalA activation was shown to be higher in a majority of mammary adenocarcinomas versus matched patient normal mammary tissue. These studies provide new insights into the importance of RalA activation in breast cancer, as well as the molecules downstream of RalA and RalB that may be responsible for mediating their effects on tumorigenesis and metastasis.