Orientador: José Arnaldo Frutuoso Roveda / Coorientador: Antonio Cesar Germano Martins / Banca: Admilson Irio Ribeiro / Banca: Gerson Araújo de Medeiros / Banca: Neli Regina Siqueira Ortega / Banca: Marcos Eduardo Ribeiro do Valle Mesquita / Resumo: Embora fundamental para diversas finalidades, a identificação de espécies arbóreas pode ser complexa e até mesmo inviável em determinadas condições, motivando o desenvolvimento de métodos assistidos por inteligência computacional. Nesse sentido, estudos têm se concentrado na avaliação de características extraídas a partir de imagens da folha e, apesar dos avanços, não são aplicáveis a espécies caducifólias em determinadas épocas do ano. Logo, o uso de características baseadas na textura em imagens do tronco poderia ser uma alternativa, mas ainda há poucos resultados reportados na literatura. Portanto, a partir da revisão de trabalhos anteriores, foram realizados experimentos para avaliar o uso de métodos de inteligência computacional no reconhecimento de padrões de textura em imagens do tronco arbóreo. Para tanto, foram consideradas espécies arbóreas caducifólias nativas da flora brasileira. As primeiras análises experimentais focaram na avaliação de padrões. Como resultado, verificou-se que a melhor capacidade de generalização é alcançada combinando o uso de estatísticas de primeira e segunda ordem. Contudo, o aumento de variáveis preditoras demandou uma abordagem capaz de lidar com informação redundante. Entre as técnicas avaliadas para essa finalidade, a análise fatorial exploratória proporcionou redução na taxa de erros durante o aprendizado de máquina e aumento da acurácia durante a validação com dados de teste. Por fim, constatando que a variabilidade natural da textura... (Resumo completo, clicar acesso eletrônico abaixo) / Doutor

No description available.

The immune response to a given cancer can profoundly influence a tumour's trajectory and response to treatment, but the ability to analyse this component of the microenvironment is still limited. To this end, a number of immune marker gene signatures have been reported which were designed to enable the profiling of the immune system from transcriptomics data from tissue and blood samples. Our initial analyses of these resources suggested that these existing signatures had a number of serious deficiencies. In this study, a co-expression based approach led to the development of a new set of immune cell marker gene signatures (ImSig). ImSig supports the quantitative and qualitative assessment of eight immune cell types in expression data generated from either blood or tissue. The utility of ImSig was validated across a wide variety of clinical datasets and compared to published signatures. Evidence is provided for the superiority of ImSig and the utility of network analysis for data deconvolution, demonstrating the ability to monitor changes in immune cell abundance and activation state. ImSig was also used to examine immune infiltration in the context of cancer classification and treatment. Patient-matched ER+ breast cancer samples before and after treatment with letrozole were analysed. Significant elevation of infiltration of macrophages and T cells on treatment was observed in responders but not in non-responders, potentially revealing a biomarker for response. ImSig was also used to study the immune infiltrate in 12 cancer types. By computing the relative abundance of immune cells in these samples, their relationship to survival was investigated. It was interesting to observe that half of the cancers showed trends towards poor prognosis with increased infiltration of immune cells. ImSig alongside the network-based framework can also be used for a more explorative analysis such as to identify biomarkers and activation or differentiation states of immune cells. Melanoma is a highly immunogenic cancer and has shown tremendous success with immune checkpoint inhibitors in a subset of patients. In chapter-6, the molecular subgrouping of melanoma was explored using a network-based approach. Despite the plethora of evidence suggesting various aspects of the immune system to contribute towards the response to immunotherapy in melanoma, there has been little to no effort to consider this heterogeneity while developing molecular subgroups. The use of ImSig was therefore explored for the stratification of melanoma patients into immuno-subgroups. The subgrouping methodology divided the tumours into four groups with different immune profiles. Interestingly, these groupings showed prognostic significance, reiterating the need to consider the heterogeneity of immune cells in future studies. On identifying the most dominant phenotypes that contribute towards prognosis of these patients and in comparison to the published subgroupings of melanoma, we argue that the subgroup of samples enriched in keratin genes are not clinically meaningful. ImSig and the associated analysis framework described in this work, support the retrospective analysis of tissue derived transcriptomics data enabling better characterisation of immune infiltrate associated with disease, and in so doing, provide a resource useful for prognosis and potentially in guiding treatment.

The advent of next-generation sequencing technologies is transforming biology by enabling individual researchers to sequence the genomes of individual organisms or cells on a massive scale. In order to realize the translational potential of this technology we will need advanced information systems to integrate and interpret this deluge of data. These systems must be capable of extracting the location and function of genes and biological features from genomic data, requiring the coordinated parallel execution of multiple bioinformatics analyses and intelligent synthesis of the results. The resulting databases must be structured to allow complex biological knowledge to be recorded in a computable way, which requires the development of logic-based knowledge structures called ontologies. To visualise and manipulate the results, new graphical interfaces and knowledge acquisition tools are required. Finally, to help understand complex disease processes, these information systems must be equipped with the capability to integrate and make inferences over multiple data sets derived from numerous sources. RESULTS: Here I describe research, design and implementation of some of the components of such a next-generation information system. I first describe the automated pipeline system used for the annotation of the Drosophila genome, and the application of this system in genomic research. This was succeeded by the development of a flexible graphoriented database system called Chado, which relies on the use of ontologies for structuring data and knowledge. I also describe research to develop, restructure and enhance a number of biological ontologies, adding a layer of logical semantics that increases the computability of these key knowledge sources. The resulting database and ontology collection can be accessed through a suite of tools. Finally I describe how the combination of genome analysis, ontology-based database representation and powerful tools can be combined in order to make inferences about genotype-phenotype relationships within and across species. CONCLUSION: The large volumes of complex data generated by high-throughput genomic and systems biology technology threatens to overwhelm us, unless we can devise better computing tools to assist us with its analysis. Ontologies are key technologies, but many existing ontologies are not interoperable or lack features that make them computable. Here I have shown how concerted ontology, tool and database development can be applied to make inferences of value to translational research.

An exome sequencing analysis pipeline was constructed to analyze NET germline and somatic samples. SNPs and INDELs were called and annotated from germline and somatic tissue. CNVs were also called for the tumor samples. This was accomplished using open source bioinformatics software that has been developed by the research community. Broad Institute "best practices" were followed. Some of the tools that were used include BWA, SAMtools, GATK, Varscan, VT, VEP, and GEMINI. Computational resources were provided by The University of Iowa NEON computer cluster. 57 germline samples and 15 tumor samples across 23 families with a history of NETs produced 4,452 germline variants, 1,695 somatic variants, 5,853 LOH events, and 627 CNV calls. False positive and driver candidacy filtering was applied. One family with Currarino syndrome has an inherited germline missense variant in MNX1. This variant has a phred-scaled Combined Annotation Dependant Depletion score of 35, putting it in the top 0.031% of deleterious variants. CNV analysis demonstrates that 8 of the 15 tumor samples have large-scale deletions of chromosome 18, three of which have nearly the entire chromosome deleted. An affected tumor suppressor gene in this region includes DCC, which is present in all three variant discovery techniques. Variant prioritization techniques are effective, but need further development to increase candidate variant/gene discovery rate.

Meiosis as a general process is prevalent across the eukaryotes, as are the orthologs of many genes encoding proteins known to function in meiosis. However, many organisms have experienced derived losses of otherwise well-conserved meiosis genes without losing meiosis and sexual reproduction. Although this general conservation of meiosis genes and precedent for derived meiosis gene losses has been previously established, questions remain about the frequency of and evolutionary forces contributing to these trends. This work sought (i) to characterize the phylogenetic distribution of 15 meiosis genes (most of which are known to function only in meiosis) in the exemplar eukaryotic kingdom Fungi and (ii) to use this dataset to investigate evolutionary processes contributing to the loss and retention of these genes. Orthologs of 15 meiosis genes (Rad51, Rad21, Spo11, Rec8, Dmc1, Hop2, Mnd1, Sae3/Swi5, Mei5/Sfr1, Pch2, Hop1, Msh4, Msh5, Mer3, Zip3) were identified by BLAST-based techniques and phylogenetically validated in most of the 109 publicly available sequenced fungal genomes investigated, but numerous putative derived losses were also detected. Rad51, Rad21, Rec8, and Spo11 were nearly universally conserved; the remaining genes were each undetectable or independently pseudogenized multiple times within fungi, particularly often for Pch2. Genes with previously known functional interactions tended to show parallel presence, absence, or pseudogenization patterns. Although this work primarily established the conserved presence of meiosis gene orthologs at the DNA level, examination of expressed sequence tags (ESTs) showed that many species--including some not previously known to undergo sexual reproduction--were competent to transcribe (and often splice) mRNA from the identified meiosis genes. Factors potentially influencing derived meiosis gene losses were investigated in two ways. First, degenerate PCR was used to amplify loci expected to contain orthologs of Msh4, Msh5, Pch2, and Zip3 in various Aspergillus species closely related to Aspergillus nidulans (a species with undetected or pseudogenized orthologs of these four genes.) The loss of Pch2 substantially predated the pseudogenization of Msh4, Msh5, and Zip3. Evolutionary rate analyses using the Ka/Ks ratio found no change in nonsynonymous substitution patterns in Msh4 and Msh5 in species that had lost Pch2 compared to those retaining Pch2. Elevated Zip3 Ka/Ks values were found in species with pseudogenized Msh4 and Msh5, suggesting possible obligate functional interactions of Zip3 with Msh4 and Msh5. Second, phylogenetically independent contrasts (PIC) analyses were performed on species from the 109-taxon inventory with published chromosome number and chromosome size estimates to investigate whether changes in either parameter were consistently associated with changes in the presence or absence of meiosis genes. Many analyses had low statistical power, neither detecting nor being able to exclude an association between gene loss and the tested variables. However, several comparisons did detect significant or nearly significant trends: for example, fungi that had lost genes related to crossover interference (Msh4, Msh5, or Pch2) tended to have fewer and/or larger chromosomes than their closest relatives without gene loss. A final objective was to determine the distribution of meiosis genes in lichenized fungi and green algae to see whether this form of symbiosis was associated with differences in the presence or molecular evolution of meiosis genes. Rad51, Dmc1, and Mnd1 were each amplified by degenerate PCR from multiple lichenized fungi that lacked sequenced genomes, and no systematic difference in evolutionary rate was found between examined lichenized fungi compared to other examined classes in phylum Ascomycota. Bioinformatic analyses of meiosis gene distribution in green algae revealed not only no obvious increased tendency for derived gene losses in examined lichenized green algae but also very few derived meiosis gene losses in green algae in general. This suggests that lichenization may not be associated with consistent differences in the evolution of meiosis genes in either fungal or green algal symbionts. The green algal results also illustrate the need to investigate the extent to which eukaryotes as a whole exhibit the same trends of meiosis gene evolution described here for fungi: frequent derived losses of meiosis genes, genes encoding proteins with function interactions showing similar distributions, likely roles for post-transcriptional regulation of meiosis gene transcripts, and loss of crossover distribution-related genes potentially being associated with constraints on chromosome size and/or haploid chromosome number.

Proteomic technology has been increasingly incorporated into agricultural research, as characterization of proteomes can provide valuable information for potential biomarkers of health and physiological status of an animal. As dairy cattle are a dominant production animal in the USA, their biofluids such as milk, blood, urine, and rumen fluid have been examined by proteomic analysis. The research outlined herein was performed to further characterize the dynamics of specific proteomes and relate them to dairy cattle physiology. The first experiment evaluated the diurnal dynamicity of the rumen metaproteome in Holstein dairy cattle. Rumen fluid was collected from three mid to late lactation multiparous dairy cattle (207 ± 53.5 days in milk) at three time points relative to their first morning offering of a total mixed ration (TMR) (0 h, 4 h, and 6 h after feeding). Samples were processed and labeled using Tandem Mass tagging before being further fractionated with a high pH reversed-phase peptide fractionation kit. Samples were analyzed by LC-MS/MS and statistically analyzed for variations across hour of sampling using the MIXED procedure of SAS with orthogonal contrasts. A total of 242 proteins were characterized across 12 microbial species, with 35 proteins identified from a variety of 9 species affected by time of collection. Translation-related proteins were correlated positively with increasing hour of sampling while more specific metabolic proteins were negatively correlated with increasing hour of sampling. Results suggest that as nutrients become more readily available, microbes shift from conversion-focused biosynthetic routes to more encompassing DNA-driven pathways. The second experiment aimed to characterize the milk fat globule membrane (MFGM) proteomes of colostrum and transition milk for comparison from multi- (n = 10) and primiparous (n = 10) Holstein dairy cattle. Samples were collected at four timepoints post-partum (milkings 1, 2, 4, and 14). After isolation of the protein lysates from the MFGM, proteins were labeled using Tandem Mass tagging and analyzed using LC-MS/MS techniques. Protein identification was completed using MASCOT and Sequest in Proteome Discoverer 2.2. Protein abundance values were scaled and analyzed using the MIXED procedure in SAS to determine the effect of parity, milking number, and parity x milking number, and the adaptive false-discovery rate (FDR)-adjusted P values were determined using the MULTTEST procedure of SAS. There were 104 proteins identified within the MFGM. Statistical analysis revealed that 44.2% of proteins were affected by parity, 70.2% by milking number, and 32.7% by the variable of parity x milking number. There was a two-fold difference in calcium sensing S100 proteins in cows differing in parity possibly due to the multiparous mammary gland being more adapted to the physiological demand of lactation or the lesser requirement of calcium in primiparous cows because of a lower production rate.

Transmembrane beta-barrels (TMBB) are a special structural class of proteins predominately found in the outer membranes of Gram-negative bacteria, mitochondria, and chloroplasts. TMBBs are surface-accessible proteins that perform a variety of functions ranging from iron acquisition to osmotic regulation. These properties make TMBBs tempting targets for vaccine or drug therapy development A prediction method based on the physicochemical properties of experimentally characterized TMBB structures was developed to predict TMBB-encoding genes from genomic databases. The algorithm's prediction efficiency was tested using a non-redundant set of sequences from proteins of known structure. The algorithm was based on the work of Wimley (2002), but was improved because of its disappointingly high false-positive prediction rate. The improved prediction algorithm developed in this study was shown to be more accurate than previously published prediction methods. Its accuracy is near 99% when using the most efficient prediction criteria, i.e. where the most known TMBBs are correctly predicted and the most non-TMBBs are correctly excluded. The improved algorithm was used to predict the abundance of TMBBs in 611 chromosomes from Gram-negative and acid-fast bacteria. The average predicted abundance of genomic TMBBs was 3%, which is consistent with previous estimates Predicted outer membrane protein L (OmpL) from Salmonella typhimurium LT2, was tested as a model for validating the prediction method. All of the physicochemical and spectroscopic properties exhibited by OmpL are consistent with other known TMBBs. Recombinant OmpL localizes to the outer membrane when expressed in Escherichia coli; has a beta-sheet-rich secondary structure with stable tertiary contacts in the presence of either detergent micelles or a lipid bilayer; OmpL also forms a pore through which small hydrophilic solutes can diffuse. Together, this data proves that OmpL is a true TMBB, which supports the computational prediction. This work significantly contributes to the advancement of TMBB research / acase@tulane.edu

Emerging topic detection algorithms have the potential to assist researchers in maintaining awareness of current trends in biomedical fields--a feat not easily achieved with existing methods. Though topic detection algorithms for news-cycles exist, several aspects of this particular area make applying them directly to scientific literature problematic. This dissertation offers a framework for emerging topic detection in biomedicine. The framework includes a novel set of weightings based on the historical importance of each topic identified. Features such as journal impact factor and funding data are used to develop a fitness score to identify which topics are likely to burst in the future. Characterization of bursts over an extended planning horizon by discipline was performed to understand what a typical burst trend looks like in this space to better understand how to identify important or emerging trends. Cluster analysis was used to create an overlapping hierarchical structure of scientific literature at the discipline level. This allows for granularity adjustment (e.g. discipline level or research area level) in emerging topic detection for different users. Using cluster analysis allows for the identification of terms that may not be included in annotated taxonomies, as they are new or not considered as relevant at the time the taxonomy was last updated. Weighting topics by historical frequency allows for better identification of bursts that are associated with less well-known areas, and therefore more surprising. The fitness score allows for the early identification of bursty terms. This framework will benefit policy makers, clinicians and researchers.

Trypanosomatids cause human diseases such as leishmaniasis and African trypanosomiasis. Trypanosomatids are protists from the order Trypanosomatida and include species of the genera Trypanosoma and Leishmania, which occupy a similar ecological niche. Both have digenic life-stages, alternating between an insect vector and a range of mammalian hosts. However, the strategies used to subvert the host immune system differ greatly as do the clinical outcome of infections between species. The genomes of both the host and the parasite instruct us about strategies the pathogens use to subvert the human immune system, and adaptations by the human host allowing us to better survive infections. We have applied unsupervised learning algorithms to aid visualization of amino acid sequence similarity and the potential for recombination events within Trypanosoma brucei's large repertoire of variant surface glycoproteins (VSGs). Methods developed here reveal five groups of VSGs within a single sequenced genome of T. brucei, indicating many likely recombination events occurring between VSGs of the same type, but not between those of different types. These tools and methods can be broadly applied to identify groups of non-coding regulatory sequences within other Trypanosomatid genomes. To aid in the detection, quantification, and species identification of leishmania DNA isolated from environmental or clinical specimens, we developed a set of quantitative-PCR primers and probes targeting a taxonomically and geographically broad spectrum of Leishmania species. This assay has been applied to DNA extracted from both human and canine hosts as well as the sand fly vector, demonstrating its flexibility and utility in a variety of research applications. Within the host genomes, fine mapping SNP analysis was performed to detect polymorphisms in a family study of subjects in a region of Northeast Brazil that is endemic for Leishmania infantum chagasi, the parasite causing visceral leishmaniasis. These studies identified associations between genetic loci and the development of visceral leishmaniasis, with a single polymorphism associated with an asymptomatic outcome after infection. The methods and results presented here have capitalized on the large amount of genomics data becoming available that will improve our understanding of both parasite and host genetics and their role in human disease.