Power system parameters forecasting using Hilbert-Huang transform and machine learning

A novel hybrid data-driven approach is developed for forecasting power system parameters with the goal of increasing the efficiency of short-term forecasting studies for non-stationary time-series. The proposed approach is based on mode decomposition and a feature analysis of initial retrospective data using the Hilbert-Huang transform and machine learning algorithms. The random forests and gradient boosting trees learning techniques were examined. The decision tree techniques were used to rank the importance of variables employed in the forecasting models. The Mean Decrease Gini index is employed as an impurity function. The resulting hybrid forecasting models employ the radial basis function neural network and support vector regression. A part from introduction and references the paper is organized as follows. The second section presents the background and the review of several approaches for short-term forecasting of power system parameters. In the third section a hybrid machine learningbased algorithm using Hilbert-Huang transform is developed for short-term forecasting of power system parameters. Fourth section describes the decision tree learning algorithms used for the issue of variables importance. Finally in section six the experimental results in the following electric power problems are presented: active power flow forecasting, electricity price forecasting and for the wind speed and direction forecasting.

Helicobacter pylori: comparative genomics and structure-function analysis of the flagellum biogenesis protein HP0958

Helicobacter pylori is a gastric pathogen which infects ~50% of the global population and can lead to the development of gastritis, gastric and duodenal ulcers and carcinoma. Genome sequencing of H. pylori revealed high levels of genetic variability; this pathogen is known for its adaptability due to mechanisms including phase variation, recombination and horizontal gene transfer. Motility is essential for efficient colonisation by H. pylori. The flagellum is a complex nanomachine which has been studied in detail in E. coli and Salmonella. In H. pylori, key differences have been identified in the regulation of flagellum biogenesis, warranting further investigation. In this study, the genomes of two H. pylori strains (CCUG 17874 and P79) were sequenced and published as draft genome sequences. Comparative studies identified the potential role of restriction modification systems and the comB locus in transformation efficiency differences between these strains. Core genome analysis of 43 H. pylori strains including 17874 and P79 defined a more refined core genome for the species than previously published. Comparative analysis of the genome sequences of strains isolated from individuals suffering from H. pylori related diseases resulted in the identification of “disease-specific” genes. Structure-function analysis of the essential motility protein HP0958 was performed to elucidate its role during flagellum assembly in H. pylori. The previously reported HP0958-FliH interaction could not be substantiated in this study and appears to be a false positive. Site-directed mutagenesis confirmed that the coiled-coil domain of HP0958 is involved in the interaction with RpoN (74-284), while the Zn-finger domain is required for direct interaction with the full length flaA mRNA transcript. Complementation of a non-motile hp0958-null derivative strain of P79 with site-directed mutant alleles of hp0958 resulted in cells producing flagellar-type extrusions from non-polar positions. Thus, HP0958 may have a novel function in spatial localisation of flagella in H. pylori

Function of the PDLIM2 protein in oncogenic Wnt signalling pathways

Aberrant regulation of the Wnt signalling pathway is a recurrent theme in cancer biology. Hyper activation due to oncogenic mutations and paracrine activity has been found in both colon cancer and breast cancer, and continues to evolve as a central mechanism in oncogenesis. PDLIM2, a cytoskeletal PDZ protein, is an IGF-1 regulated gene that is highly expressed in cancer cell lines derived from metastatic tumours. Suppression of PDLIM2 inhibits polarized cell migration, reverses the Epithelial to Mesenchymal transition (EMT) phenotype, suppresses the transcription of β-catenin target genes, and regulates gene expression of key transcription factors in EMT. This thesis investigates the mechanism by which PDLIM2 contributes to the maintenance of Wnt signalling in cancer cells. Here we show that PDLIM2 is a critical regulator of the Wnt pathway by regulating β-catenin at the adherens juctions, as also its transcriptional activity by the interaction of PDLIM2 with TCF4 at the nucleus. Evaluation of PDLIM2 in macrophages and co-culture studies with cancer cells and fibroblasts showed the influence exerted on PDLIM2 by paracrine cues. Thus, PDLIM2 integrates cytoskeleton signalling with gene expression by modulating the Wnt signalling pathway and reconciling microenvironmental cues with signals in epithelial cells. Negative correlation of mRNA and protein levels in the triple negative breast cancer cell BT549 suggests that PDLIM2 is part of a more complex mechanism that involves transcription and posttranslational modifications. GST pulldown studies and subsequent mass spectrometry analysis showed that PDLIM2 interacts with 300 proteins, with a high biological function in protein biosynthesis and Ubiquitin/proteasome pathways, including 13 E3 ligases. Overall, these data suggest that PDLIM2 has two distinct functions depending of its location. Located at the cytoplasm mediates cytoskeletal re-arrangements, whereas at the nucleus PDLIM2 acts as a signal transduction adaptor protein mediating transcription and ubiquitination of key transcription factors in cancer development.