Alzheimer’s Disease (AD) is the most prevalent form of dementia, accounting for approximately 70% of the 55 million global cases (WHO). This disease is characterized by a progressive decline in memory and cognitive functions, as well as behavioral changes and visual dysfunctions, such as deficits in perception and visual processing. Neuroimaging studies reveal atrophy in the hippocampus and neocortical regions. According to the amyloid cascade hypothesis, the beta-amyloid 1-42 oligomer (Aβ42) plays a central role in the pathology, promoting the formation of amyloid plaques and neurofibrillary tangles. These pathological changes may begin up to 20 years before the onset of the first clinical symptoms. Transgenic models suggest that an imbalance in the excitation/inhibition ratio causes episodes of neuronal silencing followed by hyperactivity, contributing to sensory disintegration and behavioral deficits.

Unlike rodents, which do not naturally develop AD, domestic cats (Felis catus) possess selective circuits and columnar cortical maps homologous to those of primates. When aged, they can develop Cognitive Dysfunction Syndrome, a condition similar to human AD, characterized by amyloid plaques, neurofibrillary tangles, and neurodegeneration in various cortical areas, including the visual cortex.

In this study, our primary goal is to introduce the primary visual cortex of domestic cats as a promising translational model for the study of AD. Next, we investigate the impact of acute exposure to Aβ42 oligomers on neuronal activity and functional connectivity in the primary visual cortex of cats. The experimental protocol involves the implantation of 4x4 microelectrode arrays (250 µm spacing) in bilateral homologs of areas 17 or 18 in anesthetized cats (n=5), with visual stimulation using moving gratings (2500 ms) in 16 directions at 2 Hz and 0.5 cycles/degree (area 17) or 0.15 cycles/degree (area 18). We analyze changes in fundamental neuronal computations, such as firing rate and orientation selectivity index. Additionally, we examine how the peptide modifies network dynamics through a short-term plasticity adaptation protocol.

This study allows precise spatial and temporal control over the action of the Aβ42 oligomer, avoiding compensatory mechanisms typical of transgenic models, in a cortex homologous to that of humans. The results aim to contribute to the understanding of the early pathophysiological mechanisms of AD.

This work explored the genetics of Major Depressive Disorder (MDD) through an integrated approach that combined multi-trait Genome-Wide Association Studies (GWAS) with functional genomics. Given the polygenic nature of depression and the challenge of "missing heritability", multi-trait analysis was employed to increase statistical power by leveraging the shared genetic architecture with other psychiatric disorders. The Multi-trait Analysis of GWAS software was utilized for this purpose, selecting phenotypes based on rigorous criteria of genetic correlation, ancestry, and sample size. After the multi-trait analysis, significant loci were annotated, and the majority of variants were found in non-coding regions of deoxyribonucleic acid, which motivated the performance of a colocalization study. For genome-wide Mendelian Randomization and colocalization, a bioinformatics pipeline called Causeway was developed, aiming to overcome the computational difficulties of existing software and ensure scalability and reproducibility with Nextflow. Colocalization analysis was performed between the multi-phenotype GWAS results and expression quantitative trait loci /protein quantitative trait loci data from blood and brain, identifying 59 significant colocalization regions, involving 31 variants and 45 distinct genes. These findings replicate previously reported genes and reveal new candidates with pharmacological potential, such as NT5C2 and ATF6B . The integration of genetic and omics data strengthens the biological reliability of the identified effector genes. The analysis also highlighted the effectiveness of colocalization in identifying potential drug targets, with 11 genes found to interact with approved or in-progress drugs, including some used for mood disorders. In summary, this work demonstrates the efficacy of multi-trait analysis and the Causeway pipeline in discovering and characterizing new genetic markers and effector genes for MDD. The findings provide valuable insights into the biology of depression, emphasizing the role of inflammation and cellular homeostasis, and open new avenues for investigating therapeutic targets and the development of more precise and personalized interventions.

Spondyloarthropathies (SpA) comprise a heterogeneous group of inflammatory, autoimmune, and chronic rheumatic diseases with genetic predisposition and possible environmental and psychological triggers. This study focuses on two main forms of SpA: ankylosing spondylitis (AS) and psoriatic arthritis (PsA), which share clinical and immunopathological manifestations and therapeutic strategies, but also have specific molecular characteristics. The main objective of this study was to investigate the dysregulation of molecular pathways and differential gene expression patterns in patients with AS and PA in order to identify genetic signatures, potential biomarkers, and biological processes that are common and distinct between the two diseases. RNA-Seq data from the Gene Expression Omnibus (GEO) repository (GSE186061, GSE117769, GSE205748, and GSE221786) were used to perform differential expression analysis, protein-protein interaction (PPI), and reconstruction of regulatory networks between transcription factors (TFs) and their target genes. The integration of these data enabled the construction of gene and regulatory networks, revealing central hubs and master regulators that possibly play key roles in the pathogenesis of both diseases. The findings contribute to the systemic understanding of spondyloarthropathies, providing insights for the identification of new therapeutic targets and the development of more accurate and individualized diagnostic strategies.

The integration of genomic data into personalized nutrition has advanced significantly, providing new insights into the influence of single nucleotide polymorphisms (SNPs) on metabolic processes and responses to nutritional interventions. In this context, this thesis is structured into two parts. In the first part, a bioinformatics platform was developed to integrate SNPs related to genes involved in metabolic pathways associated with diseases, utilizing databases such as KEGG and GeneCards. The platform enables the mapping of SNPs to metabolic pathways, diseases, and global allele frequencies, consolidating essential information for understanding the genetic impact on various diseases. In the second part, the relationship between creatine supplementation and renal function was investigated based on gene expression analysis using databases such as GTEx and GEO. Genes such as SLC6A8, IGF1, and AKT1 were analyzed under different renal conditions, including nephrosclerosis, transplant rejection, and renal carcinomas. The results highlight the relevance of bioinformatics in interpreting data available in biological databases, emphasizing the need for integration and analysis tools to enable healthcare professionals to access this information and support clinical decisions with greater precision.

Proteomics has been regarded as a promising technology, capable of providing insights into protein levels in various biological and clinical models. Proteomics has been considered a promising technology, capable of providing insights It can provide a quantitative description of the state of a biological system through the study of protein abundance profiles. Biomarkers are molecular markers found in clinical samples which may aid disease diagnosis or prognosis. High-throughput techniques allow prospecting for such signature molecules by comparing gene expression between normal and sick cells. Cancer-testis antigens (CTAs) are promising candidates for cancer biomarkers due to their limited expression to the testis in normal conditions versus their aberrant expression in various tumors. CTAs are routinely identified by transcriptomics, but a comprehensive characterization of their protein levels in different tissues is still necessary. Mass spectrometry-based proteomics allows the characterization of many cellular types and the production of large amounts of data while computational tools allow the comparison of multiple datasets, and together those may corroborate insights obtained at the transcriptomic level. Here a computational meta-analysis explores the CTAs protein abundance in the proteomic layer of healthy and tumor tissues. The combined datasets present the expression patterns of 17,200 unique proteins, including 241 known CTAs previously described at the transcriptomic level. Those were further ranked as significantly enriched in tumor tissues (23 proteins), exclusive to tumor tissues (26 proteins) or abundant in healthy tissues (8 proteins). Our study reveals the potential to enable future advancements for tumor proteome characterization and the subsequent identification of biomarker candidates and/or therapeutic targets.

The order Siluriformes, commonly known as catfishes, represents one of the largest and most diverse groups of fishes in the world. This group exhibits remarkable variation in body size across its species, ranging from just a few centimeters to giants exceeding 4 meters in length. This morphological diversity makes Siluriformes a valuable model for investigating evolutionary processes related to body size dynamics in aquatic vertebrates. Within this context, Brachyplatystoma filamentosum, known as Piraíba or Filhote, stands out as the largest catfish in the Amazon Basin and one of the largest representatives of the order. In this study, the complete mitochondrial genome (mitogenome) of B. filamentosum was sequenced and analyzed for the first time, providing novel genomic data for Amazonian ichthyofauna. The mitogenome of B. filamentosum comprises 16,566 base pairs, with a GC content of 42.21% and a D-loop region of 911 base pairs. The mitochondrial sequences encoding proteins, tRNAs, and rRNAs were incorporated into a comprehensive phylogenetic analysis including 137 additional species of Siluriformes and 10 outgroup species. Phylogenetic trees, constructed using maximum likelihood and calibrated over time, estimated the origin of the Siluriformes order at approximately 118.4 million years ago. The analyses indicated that the suborder Loricarioidei was the first to diversify, followed by Diplomystoidei and subsequently Siluroidei, which underwent a rapid evolutionary radiation around 94.1 million years ago. The reconstruction of body size evolution revealed 16 directional increases and 11 decreases in body size across the order, with no overall global trend. However, B. filamentosum exhibited a significant increase in size over 40.8 million years, with an estimated 5.65-fold growth rate, standing out as a remarkable case of gigantism within the order. The publication of the first complete mitogenome of B. filamentosum represents a significant advancement in the genetic and evolutionary knowledge of Amazonian species, especially considering the scarcity of molecular data for  many taxa in the region. This gap is particularly concerning given the ecological, economic, and conservation importance of Amazonian fishes. By integrating genomic data with phylogenetic analyses, this study contributes not only to the understanding of the evolutionary history of Siluriformes but also to the broader knowledge of an iconic species from the largest river basin on the planet.

The large volume of single nucleotide polymorphism data currently available has driven the development of methods capable of distinguishing neutral alterations from those associated with diseases such as cancer. Obtaining experimental evidence on the pathogenicity of variants is a labor-intensive, time-consuming, and costly process. Several in silico tools have been employed for pathogenicity prediction, including PolyPhen-2, PROVEAN, SIFT, FATHMM, MutationTaster, MutationAssessor, and LRT, as well as ensemble-based methods that combine multiple independent predictors, such as ClinPred, MetaLR, and MetaSVM. However, most of these approaches rely primarily on genomic information and allele frequency data. In recent decades, tools that integrate topological features from residue interaction networks (RINs) with outputs from conventional predictors have demonstrated superior performance. The objective of this work was to develop a classification model capable of assessing the impact of structural and topological RIN features on improving the accuracy of mutation classifiers. To this end, curated databases were constructed containing functional predictions, genomic, structural, and functional information associated with 33 cancer types, followed by the application and evaluation of several supervised machine learning algorithms. The results showed that integrating structural and topological parameters derived from RINs enhances the predictive performance of machine learning models in classifying cancer-associated missense mutations. The XGBoost-based model achieved consistent performance, with an accuracy of 74.0%, sensitivity of 73.9%, specificity of 74.1%, and an F1-score of 74.5%. These findings indicate that the proposed model presents a well-balanced trade-off between sensitivity and specificity, avoids bias toward either class, and demonstrates strong generalization capability in a highly heterogeneous scenario comprising multiple genes and distinct tumor contexts.

The explosion of genomic data in recent decades has presented a substantial challenge, requiring new approaches for efficient analysis and interpretation. This research emerges in this context, offering comprehensive bioinformatics analysis, exploring various facets of genomics and its relevance to health. The study encompasses the analysis of mitochondrial genomes of Amazonian species, the investigation of genetic variants and their correlation with the survival of gastric cancer patients in Natal-RN, and the development of the DTreePred application, designed to predict the pathogenicity of these variants. Additionally, the results of the analysis of gastric cancer patients in Belém-PA are discussed, employing machine learning for disease detection based on genetic variants. To validate the AI models developed based on the Pará population, public samples of Korean patients with and without gastric cancer were used. It is noteworthy that the most effective models achieved an accuracy of over 90% in classifying Korean patients as normal or cancer patients. This research thus highlights the productive integration of bioinformatics techniques in genomic research and the understanding of complex diseases, representing significant advances in the fields of health and genomics.

Enterococcus casseliflavus, a commonly motile, yellow-colored bacterium, is a commensal member of the gastrointestinal tract. It is occasionally found in cases of bacteremia and other human infections. One concern is that all strains of this species have the vanC gene cluster on their chromosome, which confers resistance to vancomycin. The classification of E. casseliflavus is challenging, as it presents 99% identity in the 16S analysis with E. gallinarum and, mainly, with E. flavescens, being often classified as a single species. The goal of this study was to investigate the taxonomy of E. casseliflavus and other related species by analyzing genomic data available in public databases. For this purpose, 155 genomes of species related to E. casseliflavus (E. casseliflavus, E. flavescens, E. entomosocium and E. innesii) were rescued and subjected to Average Nucleotide Identity (ANI), pangenome and phylogenomic analysis. The approaches showed three well-defined groups corresponding to three species of Enterococcus (E. casseliflavus, E. flavescens and E. innesii). With characteristics of an open pangenome, the group showed great conservation of core genes and high accessory genome diversity. Here we suggest removing the synonymous species status between the species E. flavescens and E. casseliflavus and adding the synonymous species status between E. entomosocium and E. casseliflavus.

Network visualization is a critical step for effective communication in various fields of knowledge, especially in life sciences. Currently, a gap separates network manipulation from network visualization in programming environments. Users constantly face the inconvenience of exporting network data to be laid out in external software, like Cytoscape and Gephi. We propose easylayout, a package that smoothly bridges manipulation and visualization by taking advantage of existing literate programming tools (e.g., RStudio for RMarkdown, Jupyter Notebooks). The easylayout package receives an igraph object and serializes it into a web application integrated with the RStudio interface through a Shiny server. This web application provides an environment to lay out the network by simulating attraction and repulsion forces. There is an editing mode that allows users to move and rotate vertices. The development of easylayout aims for computational performance, so that even low-end devices are able to work with networks with thousands of vertices and tens of thousands of edges. One way this is done is by curating high-performing graph drawing algorithms like VivaGraphJS (github.com/anvaka/VivaGraphJS) and Cosmos (github.com/cosmograph-org/cosmos). Once the user finishes tinkering the layout, it is sent back to the parent environment to be visualized through popular libraries like ggplot2 and matplotlib. Although the current implementation of easylayout focuses on refining the experience in the RStudio ecosystem, the use of web technologies make it easily portable to similar environments, like Jupyter Notebooks and VSCode. We expect this tool not only to significantly reduce the time spent laying out networks, but also to allow researches to generate more compelling figures.

Cell type identification is a critical step in the computational analysis of scRNA-Seq experiments, involving the unsupervised grouping of cells based on gene expression profiles. Traditional methods relying on canonical gene markers exhibit limitations, such as sensitivity to variations and the absence of characteristic genes for certain cell types. To address these challenges, we propose a novel approach combining machine learning algorithms with feature selection. Our methodology involves selecting a dataset suitable for training a model to ensure generalization to new data. We chose a comprehensive dataset encompassing the central and peripheral nervous system from mice at different developmental stages. Subsequently, feature selection was applied using the DUBStepR algorithm, considering gene-gene correlations to identify optimal features for cell classification. The resulting dataset, composed of 28,795 cells and 16,960 genes, was used to train and evaluate models employing k Nearest Neighborhood (kNN), Decision Tree (DT), Naive Bayes (NB), Support Vector Machine (SVM) and Multilayer Perceptron (MLP) algorithms. All models demonstrated F1-scores exceeding 90%, except for NB. Testing on a human brain scRNA-Seq dataset confirmed the robustness of the algorithms, with area under curve (AUC) values indicating accurate cell classification. SVM and MLP were selected for further analysis due to lower false positive and false negative rates. Comparisons with existing tools such as scAnnotatR and ACTINN highlight the versatility of our approach, particularly when dealing with diverse cell types. Next, we applied the SVM and MLP models to classify neurons generated in vitro human-induced neurons (hiNs) generated using distinct protocols, achieving consistent results in identifying glutamatergic and GABAergic neurons. We also attempted to classify hiNs according to cells of different brain regions, revealing challenges in classifying GABAergic neurons by region, possibly due to a limited number of optimal features. Gene expression analysis and Gene Set Enrichment Analysis (GSEA) contributed to identify gene sets associated with the electrophysiological maturation of glutamatergic hiNs generated through an alternative protocol using ASCL1 compared to other protocols. Regulatory network analysis identified master transcription factors with higher activity specifically in this protocol. In conclusion, our integrated approach of feature selection and machine learning algorithms offers an alternative way of identifying cell groups based on gene expression profiles, enhancing the refinement of single-cell analysis in the context of differential gene expression, GSEA, and regulatory gene networks.

Methods for reconstructing evolutionary scenarios are important tools that help to better understand biological systems under the perspective of its origins and evolution. The primary concept for understanding them lies in the relationships established by comparing genomes from different species to form gene families known as Orthologous Groups (OGs). Orthologs are genes from distinct species derived from their last common ancestor (LCA), tipically having similar functions in each organism. By observing the phyletic pattern of an OG in a species tree, it is possible to calculate the LCA where most probably the trait represented by the OG emerged. Although this process can be trivial when applied to a single gene, it remains chalenging for large-scale queries. The bridge algorithm, structured as a R software package, allows to interrogate several hundreds to thousands of OGs at once, assigining evolutionary roots to each OG. This thesis constitutes a comprehensive reference to the method of rooting orthologous genes employed by the bridge algorithm, presenting detailed logic, implementation, accuracy, and performance.

Since December 2019, the global impact of the COVID-19 pandemic, caused by the SARS-CoV-2 virus, has been profound. Early identification of the virus’s taxonomic classification and genomic origin is critical for strategic planning, containment, and treatment. Deep learning techniques have proven successful in addressing various viral classification challenges, including diagnosis, metagenomics, phylogenetics, and genomic analysis. Motivated by these advances, this study introduces an effective viral genome classifier for SARS-CoV-2, utilizing a convolutional neural network (CNN) framework. This research employed image representations of complete genome sequences to train the CNN, leveraging two distinct datasets: one based on k-mer image representation and the other on Chaos Game Representation (CGR). The k-mer dataset was used for taxonomic classification experiments of the SARS-CoV-2 virus, while the CGR dataset focused on classifying variants of concern (VOC) of SARS-CoV-2. The CNN achieved remarkable performance in taxonomic classification, with accuracy rates ranging from 92% to 100% on the validation set and between 98.9% and 100% on the test set containing SARS-CoV-2 samples. These results demonstrate the model’s adaptability for classifying other emerging viruses. For the classification of SARS-CoV-2 variants using CGR images, the CNN delivered even higher accuracy, reaching 99.9% on the validation set and 99.8% on the test set. The findings underscore the applicability of deep learning techniques in genome classification tasks, providing a robust tool for the early detection and classification of viral threats. The integration of CNNs with k-mer and CGR image representations presents a novel and effective method for viral genome analysis, supporting ongoing efforts in virology and public health.

Leptospirosis is considered a globally important zoonosis due to its widespread distribution and virulence, affecting both humans and commercially important animals. It is caused by pathogenic bacteria of the genus Leptospira and phylum Spirochaetes, and contamination occurs through direct or indirect contact with the contaminant agent present in the environment, such as urine from infected animals or contaminated water and soil. The genus has 68 species that can be grouped into two major groups according to their lifestyle: pathogenic and saprophytic. In addition to taxonomic classification, samples of these genera can be classified based on their antigenic characteristics into serogroups and serovars. Serological classification is of great relevance in the fields of epidemiology and clinical analysis, but the methods used for this classification are laborious, require infrastructure and specialized labor, and take days to obtain results. In this study, we aimed to find genetic patterns associated with the serological classification of Leptospira bacteria by analyzing the genetic composition of the rfb locus and proposing methods that allow for the classification of Leptospira samples at the serogroup level. To do this, we used genomic data from 68 species classified into 27 serogroups, which are distributed in 722 samples available in public databases. We identified the genes that are part of the rfb locus through orthologous groups in samples that contained the intact rfb locus in a single contig. We used a hierarchical clustering method to group samples with similar genetic profiles of the rfb locus. This analysis made it possible to contemplate the diversity of the genetic composition profile of the rfb locus in the genus Leptospira and to observe correspondence between serogroup classification and the groups formed by hierarchical clustering. The generated clustering suggests the classification of samples into six large classes that, in addition to presenting serological affinity, share similarities in the genetic composition of the rfb locus. It was observed that samples of the same serogroup share similarities in the genetic composition of the rfb locus. Additionally, it was possible to verify the existence of different gene blocks that may be conserved in samples belonging to different species and serogroups. It is presumed that different combinations of these gene blocks result in the synthesis of different O-antigen structures of lipopolysaccharides and consequently different serogroups. This study allows for the suggestion of molecular markers that allow for the use of molecular strategies for the serological identification of Leptospira.

Medulloblastoma (MB) is one of the most common pediatric brain tumors and it is estimated that one-third of patients will die from the disease. The lack of accurateprognostic biomarkers is a major challenge for the clinical improvement of thosepatients, with conventional prognostic parameters having limited and unreliable correlations with the disease outcome. Acknowledging this issue, our aim was to build a gene signature and evaluate its potential as a new prognostic model for patients with the disease. Hypermethylation of tumor suppressor genes and hypomethylation of oncogenes are methylation dysregulations crucial for cancer tumorigenesis and tumor maintenance, and it is no exception for MB. In this study, we used six datasets totaling 1679 MB samples, including RNA gene expression and DNA methylation data from primary MB as well as control samples from healthy cerebellum. We identified methylation-driven genes (MDGs) in MB, genes whose expression is correlated with their methylation and which are also differentially methylated in relation to healthy tissue. After, LASSO regression, a supervised machine learning statistical method, was used with the MDGs as a parameter resulting in a two-gene signature (GS-2) of candidate prognostic biomarkers for MB (CEMIP and  NCBP3). Using a risk score model, we confirmed the GS-2 impact on overall survival (OS) with Kaplan-Meier analysis (log-rank p < 0.01). We evaluated its robustness and accuracy with receiver operating characteristic (ROC) curves predicting OS at 1, 3 and 5 years in multiple datasets (training set: 77.2%, 73.2% and 71.2%, mean in three validation sets: 83.6%, 77.6%, 75.4% at 1, 3 and 5 years respectively). We evaluated GS-2 as an independent prognostic biomarker with multivariable Cox regression which showed p-value < 0.01 in all four datasets evaluated. The methylation-regulated GS-2 risk score model can effectively classify patients with MB into high and low-risk, reinforcing the importance of this epigenetic modification in the disease. Such genes stand out as promising prognostic biomarkers with potential application for MB treatment.

Guillain-Barré Syndrome (GBS) and multiple sclerosis are autoimmune diseases associated with an immune response against peripheral (PNS) and central nervous system (CNS) autoantigens, respectively. Most studies on GBS immunopathology investigate the cross-reactivity between myelin sheath ganglioside antigens and carbohydrates from Campylobacter jejuni bacteria. However, GBS has a spectrum of subtypes and, particularly, the Acute Inflammatory Demyelinating Polyradiculoneuropathy (AIDP) form has little evidence of a relationship with C. jejuni or of autoimmunity against gangliosides. The immunopathology of multiple sclerosis is better understood, with several protein autoantigens reported in the literature. In the present work, we screened the databases “The Human Protein Atlas,” AFND, and IEDB to select abundant proteins from the human nervous system (HNS), immunogenic proteins of the Epstein-Barr virus, and HLA haplotypes, respectively. Then we constructed a pipeline with several open-source computational tools to predict HLA binding to peptides and cytokine production. The following analysis used ten proteins from the HNS and 28 from EBV to predict the binding peptides of 21 common HLAs in the world population. From the search for haplotypes in the AFND, we found 1359 registered haplotypes distributed among 51 pairs of HLAs. After that, our pipeline compared nonapeptide anchors of EBV and myelin proteins for identity at critical residues for interaction with the T-cell receptor (TCR), establishing three selection criteria according to the relevance of each contact for TCR-peptide-MHC interaction. According to these criteria, all nervous system proteins presented peptides with relevant identity with EBV peptides. The prediction of IL-4 or IFN-γ cytokine stimulation allowed the discovery of which pairs of similar nonamers can induce the activation of Th1 or Th2 cells and perhaps cause autoimmunity through molecular mimicry. Seven proteins (APLP1, CNP, GlialCam, MAG, MBP, Periaxin, and PLP) presented pairs of similar peptide stimulators of cytokines IL-4, IFN-γ, or both. The P0 protein also presented pairs capable of inducing IL-4 or IFN-γ, though restricted to one or few HLAs or haplotypes. Given the high number of possible peptides that can cause molecular mimicry, our results align with the hypothesis that multiple antigens can cause immunity in multiple sclerosis and GBS. The nonamer pairs found here support further experimental investigations of these autoantigens and contribute to a better understanding of both pathologies.

Bioactive compounds have been studied in order to offer better efficacy and selectivity against various diseases, representing a promising scenario in drug development. Recently, a series of acridinic derivatives was synthesized and exhibited antileishmanial and anticancer activity. However, the concept of "one target, one drug, one disease" is not always true, as compounds with previously described therapeutic applications can act on more than one target. Based on this, this work aimed to identify, through reverse virtual screening based on the receptor, the probable mechanism of action of spiro-acridinic derivatives. Additionally, the mechanism of action was confirmed through in vitro enzymatic assays. Using these approaches, Chapter I of this work presents the identification, through computational methodologies, of the pteridine reductase 1 (PTR1) enzyme of L. major as a potential target for spiro-acridinic compounds. Additionally, we found the chitinase B1 (CHIB1) enzyme of Aspergillus fumigatus as a potential target against Aspergillosis. For PTR1, docking and molecular dynamics assays presented the high stability of compound 1 in the active site of the enzyme. For CHIB1, other derivatives were subjected to molecular docking and molecular dynamics, identifying 3 compounds with the best profile for the target. In Chapter II, in vitro assays were performed to experimentally confirm the action of spiro-acridinic derivatives on the studied enzymes. For PTR1, in vitro assays demonstrated a KD of 33.1 μM for the best compound, while for chitinase, the best compound showed an IC50 of 0.6 ng/μL. Therefore, this work demonstrated the high efficiency of reverse virtual screening as a target prediction approach. Additionally, the program allowed for characterizing its potency, inhibition modality, and interaction profile with its therapeutic target. Thus, spiro-acridinic derivatives can act as multi-target inhibitors of Leishmania's PTR1 and fungal chitinase.

Ewing’s Sarcoma (ES) is a highly aggressive disease and the second most frequent pediatric bone neoplasm. The ES hallmark is the presence of the aberrant transcription fator EWSR1-FLI that drives metabolic reprogramming in ES. The ES survival rate has increased at the cost of high toxicity that limits survival rates and causes significant morbidity. Therefore it is crucial to identify and obtain a complete understanding of the pathways that impact ES survival for development of novel diagnostics and therapeutic strategies. Here, we identified differences at the transitional level between ES patients with short-term survivors (STS) and long-term survivors (LTS) based on transcriptional data available in three public datasets, applying the transcriptogram analysis. Three differentially expressed clusters commons across the cohorts analyzed were identified. Processes related to DNA damage response and repair, immune response, apoptosis and autophagy were dysregulated between the STS and LTS groups. Furthermore, the functional enrichment of the common genes between three clusters and ES regulons highlight the upregulation of the Hippo pathway in STS patients. Our analysis suggests that different processes may be guiding the outcome of ES patients in an integrated way and may contribute to the diversity of phenotypes driven by the EWSR1-FLI1 expression fluctuation.

About 71% of the Earth's surface ecosystem is covered by ocean, responsible for containing 97% of all the Earth's water, with plankton being its dominant life form. Microorganisms play an essential role in maintaining the planet's system, since they are sources of energy and nutrients for living beings, realize almost half of net primary production, maintain chemical balance in the atmosphere and export photosynthetically fixed carbon to the deepest layers of the ocean. Ocean ecosystem biology studies how biotic and abiotic factors determine ocean ecosystem properties. With the emergence of global-scale studies of open sea organisms, much has been discovered about the genomics of oceanic microbial communities. One of the main projects in this category today is Tara Oceans, a multidisciplinary project aiming to understand the diversity, interactions, functions and phenotypic complexity at taxonomic and spatial scales of plankton, through about 35000 samples with millions of organisms collected at depths of up to 2000 m in 210 stations across the oceans. The main objective of this work was to compare oceanic samples exploring the abundance, diversity, and function of microorganisms found in tropical and subtropical zones, in order to understand how these factors affect the biodiversity of these species. For this, only samples from collection stations that presented samples in the three depth layers (SRF, DCM and MES) simultaneously were selected. This filtering resulted in 8 stations with a sum of 76 samples. These data were processed through the MEDUSA pipeline, following the standard flow of a metagenomic analysis: pre-processing, sequence alignment with a reference protein bank, taxonomic classification and functional annotation. With the results obtained from this process, it was possible to compare the abundance and diversity of these samples and to note and analyze the results of this comparison. A greater diversity of organisms was observed in the deepest layer (MES) and there was no significant difference in abundance between the explored depth layers. Some biological functions are unique to each depth layer, indicating its particularities and functional diversity. No significant distinction of abundance, diversity or function was observed comparing exclusively samples from collection stations in different geographic points.

This thesis presents three studies carried out in the sphere of molecular modeling based on principles of Quantum Mechanics. Additionally, molecular evolution methods complemented some results. The first study portrays the particularities of the performance of the energy and computational cost results of 9 combinations of models based on DFT (DFT -- Density Functional Theory) in an organometallic system formed by the divalent zinc cation and the enzyme Porphobilinogen Synthase PBGS. The interaction energies were obtained using the Fragmentation with Conjugated Caps (MFCC) scheme. The results of the total interaction energy profile showed linear quantitative differences, but were qualitatively uniform. The computational processing time dependency is more associated with the choice of basis set than the exchange and correlation functional. The second study presents a biochemical description from the interaction energy results obtained in the previous study, analyzing the biochemical profile of the most relevant PBGS residues that interact with zinc. In addition, a phylogenetic and cluster analysis was performed that evaluated the conservation of the relevant amino acids identified in the zinc-PBGS system. The most important intermolecular interactions were due to the participation of amino acids CS0122, CIS0124, CIS0132, ASP0169, SER0168, ARG0221, HIS0131, ASP0120, GLY0133, VAL0121, ARG0209, and ARG0174. Among these residues, ASP0120, GLI0133, HIS0131, SER0168, and ARG0209 stood out for occurring in all groups generated by the unsupervised cluster analysis. On the other hand, triple cysteines at 2.5 Å of zinc (CIS0122, CIS0124, and CIS0132) showed the highest attraction energy and are absent in Viridiplantae, Sar, Rhodophyta, and in some groups of Bacteria. The third work presented here investigates the interactions between the Lys49-PLA 2 toxin from the venom of Bothrops moojeni, which causes tissue necrosis in snakebite victims, and two compounds (varespladib, aspirin) with the potential to inhibit the myotoxic activity of these proteins. The methodology utilized here also uses quantum methods based on DFT within the MFCC scheme. From this study, it was possible to predict the relevance of the amino acids that form the Lys49-PLA 2 binding site, among them, we can mention LIS0069, LIS0049, LEU0005, ILE0009, CIS0029, GLI0030, HIS0048, PRO0018, ALA0019, CIS0045, TIR0052, TIR0022, PRO0125*, and FEN0126* which anchor varespladib and residues LIS0069, LIS0049, GLI0032, LEU0002, and LEU0005 which anchor aspirin.

Patients with clear cell renal cell carcinoma (ccRCC) have poor survival outcomes, especially if it has metastasized. It is of paramount importance to identify biomarkers in genomic data that could help predict the aggressiveness of ccRCC and its resistance to drugs. Thus, we conducted a study with the aims of evaluating gene signatures and proposing a novel one with higher predictive power and generalization in comparison to the former signatures. Using ccRCC cohorts of the Cancer Genome Atlas (TCGA-KIRC) and International Cancer Genome Consortium (ICGC-RECA), we evaluated linear survival models of Cox regression with 14 signatures and six methods of feature selection, and performed functional analysis and differential gene expression approaches. In this study, we established a 13-gene signature (AR, AL353637.1, DPP6, FOXJ1, GNB3, HHLA2, IL4, LIMCH1, LINC01732, OTX1, SAA1, SEMA3G, ZIC2) whose expression levels are able to predict distinct outcomes of patients with ccRCC. Moreover, we performed a comparison between our signature and others from the literature. The best-performing gene signature was achieved using the ensemble method Min-Redundancy and Max-Relevance (mRMR). This signature comprises unique features in comparison to the others, such as generalization through different cohorts and being functionally enriched in significant pathways: Urothelial Carcinoma, Chronic Kidney disease, and Transitional cell carcinoma, Nephrolithiasis. From the 13 genes in our signature, eight are known to be correlated with ccRCC patient survival and four are immune-related. Our model showed a performance of 0.82 using the Receiver Operator Characteristic (ROC) Area Under Curve (AUC) metric and it generalized well between the cohorts. Our findings revealed two clusters of genes with high expression (SAA1, OTX1, ZIC2, LINC01732, GNB3 and IL4) and low expression (AL353637.1, AR, HHLA2, LIMCH1, SEMA3G, DPP6, and FOXJ1) which are both correlated with poor prognosis. This signature can potentially be used in clinical practice to support patient treatment care and follow-up.

Long non-coding RNAs (lncRNAs) comprise the most representative transcriptional units of the mammalian genome, and they’re associated with organ development that can be associated with the emergence of diseases, such as cardiovascular diseases. The World Health Organization (WHO), for example, has published that cardiovascular diseases are responsible for the death of 17.9 million people each year, corresponding to 31% of all deaths all around the world. Therefore, a combination of transcripts from Gencode (M20), Ensembl (GRCm38.95) and Amaral et al. (2018) databases was used to define the set of non-redundant reference lncRNAs; and Gencode (M20) for the reference coding transcripts. In addition, bioinformatics approaches, machine learning algorithms and statistical techniques were used to define lncRNAs involved in mammalian cardiac development in a single-cell perspective. For this, the single-cell database published by DeLaughter et al. (2016) was used, in which there were data from 4 embryonic stages (E9.5, E11.5, E14.5, E18.5) and 4 post -natals (P0, P3, P7, P21) of the mus musculus model organism. Our study identified 8 distinct cell types, novel marker transcripts (coding/lncRNAs) and also, differential expression and functional enrichment analysis revealed cardiomyocyte subpopulations associated with cardiac function; meanwhile modular co-expression analysis reveals cell-specific functional insights for lncRNAs during myocardial development, including a potential association with key genes related to disease and the “fetal gene program”. Our results evidence the role of particular lncRNAs in heart development, and highlights the usage of co-expression modular approaches in the cell-type functional definition. As future work, we intend to identify the functional roles of these RNAs in the development of cardiac tissues and in cardiovascular diseases using experimental validation approaches.

The proteomic approach allows large-scale studies of protein expression in different tissues and body fluids, aiming to identify and quantify the total protein content. In the proteomic analysis process, protein identification still presents limitations despite major advances in the area. Frequently, a mass spectrometer is used to generate mass/charge values of the samples. After this process, a reference protein database (eg, UniProt) is usually used to identify proteins. However, using a reference database limits the analysis of the identification of the proteins, since it does not contain the variations in the DNA that can impact the sequence of amino acids, causing incorrect identification or making the process impossible. In this context, there are several custom databases that incorporate such genetic variations. Although they present good results, they are also limited due to the absence of some mutations, becoming another problem in the identification process. A proteogenomics database (dbPepVar) created here combines genetic variation information from dbSNP with protein sequences from NCBI's RefSeq. Public mass spectrometry datasets were used to perform a pan-cancer analysis (Ovarian, Colorectal, Breast, and Prostate), allowing the identification of unique genetic variations. In total 3,726 variant peptides were identified in ovarian cancer samples, 2,543 in prostate, 2,661 in breast and 2,411 in colon-rectal cancer. A mutational frequency analysis showed genes involved in tumor progression processes, sensitivity to chemotherapy, and risk of susceptibility to cancer. Interestingly, in many samples, C-terminal peptides from shortened proteins originating from premature termination codon (PTC) events were identified. This indicates that such proteins had escaped Nonsense-mediated decay (NMD) and, not surprisingly, NMD machinery genes are also mutated in the same samples. This suggests that the vestige of the truncated transcript may be associated with NMD machinery inefficiency caused by gene mutations. In perspective, the web portal developed as well as the analysis performed may direct studies to identify new therapeutic targets for different cancer, and one can also use our database for characterization of variants in samples of unknown genetic background, such as archived samples. The portal is available in: https://bioinfo.imd.ufrn.br/dbPepVar/

Non-coding RNAs are molecules that play decisive roles in several types of gene regulation. Identifying them is necessary for understanding the genetics of a species. Several factors, such as: low level of expression, the broad spectrum of subtypes, diverse attributes, heterogeneous functions and absence of homology between species; make the detection of ncRNAs genes a challenge. The latest bioinformatics strategies for detecting ncRNA genes have tried to identify their locations in the genomes and their secondary structures, using covariance models and artificial intelligence. The co-expression of these genes has been computationally analyzed in order to reveal their functional annotations. However, there is no consensus on which metrics and parameters to use in the process of predicting the functions of these molecules. In organisms little known, such as Arapaima gigas, the lack of reference information increases the difficulty. Additionally, even for known long non-coding RNAs, there is little functional information, which makes it difficult to explain the roles of these genes. In this work, we describe a software for discovering the non-coding genes, including their diverse types, and their functions in eukaryotic genomes. It was validated by annotating a model species (Mus musculus) and then used to explore the landscape of ncRNA in Arapaima gigas. Comparing the similarity between the functions of co- expressed genes allowed us to define confidence levels for the metrics that measure co-expression, and thus, develop a pipeline for predicting lncRNA functions, which includes metrics for non-linear correlations. The described software suite made 63307 non-coding annotations in A. gigas, including 11 types of ncRNA and 4 types of cis-regulatory regions. Of these annotations, only 706 are similar to ncRNAs already known in other species and the remaining were never described before. The exploratory analysis of lncRNA also revealed 19854 tissue specific lncRNAs and 256 lncRNAs ubiquitously expressed. Predicting the functions of these molecules revealed RNAs involved in skin pigmentation, sex differentiation, growth and defense against tumors.

The incidence of neurodegenerative diseases leading to impairment of cognitive functions and dementia have increased in recent years, mainly because of enhanced longevity in the population worldwide. Understanding the onset and progression of these pathologies can help to develop preventive and disease-modifying treatments for these diseases. In this work, using RNA-seq data obtained from two brain regions (temporal cortex and cerebellum) of human patients diagnosed with neurodegenerative diseases (Alzheimer or Progressive Supranuclear Palsy) and two animal models, 5XFAD of amyloidopathy and TauD35 of tauopathy, we performed an integrative analysis at the gene/transcript level combined with a co- expression analysis to identify similarities and discrepancies in the biological processes affected by these two diseases. So that we could compare the different data, we used the only common variable in all datasets: age of death. Thus, we divided the human data into 3 groups: A (70-80), B (81-89) and C (90+); and animals in groups of 4 months, 12 months, 17 months and 18 months. The results of the transcriptional analysis showed that gene expression alterations associated with immune-inflammatory alterations are present in AD only in the temporal cortex and not in the cerebellum, and that alteration related to synaptic transmission occurs late (groups B and C), and are found only when we use genes with isoform switches in the analysis of functional enrichment in conjunction with differentially expressed genes. In PSP, all changes associated with immune-inflammatory responses and synaptic transmission are found exclusively in temporal cortex data; however, all changes are specific for group A. In animal models, changes in 5XFAD are similar to those found in AD human brains, with gene expression alterations associated with the immune-inflammatory response present early (4 months) and synaptic terms only at late pathological stages (18 months). In TauD35 mice, this pattern is inverted, with gene expression changes associated with immune- inflammatory response identified only late (17-month group), whereas those associated with synapses could be identified early (4-month group). In addition to these results, we observed that changes in isoforms (gDTUS) are present almost exclusively in humans, and especially in AD. To refine our results, we used a co-expression approach and identified modules with specific expression and gene signatures. In AD, modules involving synapses did not differ from control, however, modules related to immune-inflammatory response, extracellular matrix and growth factor response were more active in individuals with AD. In PSP, modules with synaptic activity showed greater activity compared to control, while those related to immune response had a lower activity. To confirm the genetic identity of these modules, we also mappedmodule-specific genes to different cell types of the brain using single-cell RNA-seq data. This analysis revealed a correspondence between modules related to the immune-inflammatory response with microglial cells and, to a lesser extent in AD, astrocytes, synaptic cells with glutamatergic neurons and myelination with oligodendrocytes. Finally, we show that genes identified as risk factors for AD or PSP are present in specific co-expression. Together, these results suggest that in the amyloidopathy model and in AD, alterations in synaptic signaling form a positive feedback with the immune inflammatory response, the latter being the first; while in the model of tauopathy and PSP, the effects on inflammation are secondary to synaptic changes.

The first wave of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic hit almost all cities in Brazil in early 2020 and lasted for several months. Despite the effort of local state and municipal governments, an inhomogeneous nationwide response resulted in a death toll among the highest recorded globally. To evaluate the impact of the nonpharmaceutical governmental interventions applied by different cities - such as the closure of schools and business in general - in the evolution and epidemic spread of SARS-CoV-2, we constructed a full-sized agent-based epidemiological model adjusted to the singularities of single cities. The model incorporates detailed demographic information, mobility networks segregated by economic segments, and restricting bills enacted during the pandemic period. As a case study, we analyzed how the City of Natal - a midsized state capital - reacted to the pandemic. Although our results indicate that the governmental response was suboptimal, the restrictive mobility acts saved many lives, our simulations showed that the suspension of school activities was essential to avoid a high number of deaths (the increase would be around 525.93%). The authentic closing of Work activities would decrease the number of deaths by approximately 67.54% and religious activities by 26.7%. The absence of intervention would result in a catastrophic scenario of 6779 deaths, this number corresponds to about 0.77% of the Natal city population. The simulations show that a compartmental analysis of the alternative scenarios can inform policymakers about the most impactful measures for further surges of the pandemic and support future decisions as the pandemic progresses.

Large-scale messenger RNA sequencing (RNAseq) allows the evaluation of the diversity of transcripts expressed at a given moment in a biological system. Through bioinformatics, we can analyze the sequencing data to obtain quantitative information about gene expression, such as the differential expression of genes and their isoforms (alternative splices). In this thesis, we present two independent studies that used bioinformatics to obtain relevant information about different biological phenomena. In the first case, we used mRNA sequencing data in the brains of patients with Alzheimer's disease to study the differential expression of genes and transcripts associated with the progression of this disease. We have shown that the analysis of transcripts allows the identification of genetic changes ignored in previous studies by evaluating only the global expression of genes. Using single cell mRNA sequencing data (scRNAseq), we also map changes in gene expression in the brain of patients with Alzheimer's disease to specific cell types. The results of this first work contribute to a better understanding of the pathophysiology of Alzheimer's disease and pinpoints possible cell-type specific molecular mechanisms of the disease. In the second work developed in this thesis, we used the scRNAseq technique to study the diversity of progenitor cells in the early stages of the development of the neocortex. Through analysis of differential gene expression and the use of an approach using gene regulatory networks, we identified the transcription factor Sox9 as a master regulator of the behavior of different subtypes of neural progenitors. Confirming these findings from bioinformatics, genetic experiments to manipulate Sox9 expression levels in neural progenitors demonstrated the importance of this transcription factor in the regulation of cell proliferation and differentiation. Together, the results of this thesis demonstrate the importance of transcriptomic analysis through complementary methods for a better identification of relevant gene expression changes in different biological contexts.

Sepsis is a acute inflammatory syndrome. Accountable for most obits in ICUs all over the world. Due to its multifactorial nature, there are few studies related to gene expression regulation in pediatric septic patients. Understanding the regulatory mechanisms of sepsis could help against sepsis and also help identify key points of signaling pathways responsible for disease progression. A good strategy to identify regulatory targets of a given disease is by reconstructing its regulatory network, as well as identify its possible master regulators. Given the lack of pediatric sepsis data and the huge difference between adult and pediatric immune response, the objective of this work is to reconstruct sepsis regulatory network and identify its putative master regulators. In summary, we found 15 transcription factors that have good chance of acting as master regulators in pediatric sepsis. Specially MEF2A, TRIM25 and RFX2 were identified upregulated in septic patients in comparison to healthy individuals. Each one of them have a distinct role, that was not directly related to sepsis. But, taken together, we hypothesize that they might act together to influenciate the disease prognosis. Results herein found points towards this three transcription factors as putative master regulators of pediatric sepsis. In vitro validation of the results found in silico could shed light in the different aspects of regulatory understanding of pediatric sepsis.

Analysis of variants in clinical context and the support for the development of therapies against viral diseases are two areas which several research have used processes of integration and analysis of omics data. Assessing whether a given variant has a pathogenic impact is a challenge in the analysis of variants, especially when different tools for predicting pathogenicity point to divergent results. Regarding the development of RNA interference-based therapies, it is observed that there is a continuing need to design and evaluate the efficiency of new small-interfering RNAs (siRNAs) for each new virus that arises, like SARS-CoV-2, responsible for the COVID-19 pandemic. In this sense, it is argued in this thesis, based on the discussion of two works, that data integration and feature selection processes can contribute to the resolution of issues related to the identification of pathogenicity of variants and, in a second moment, to the availability of information and characteristics of sequences that may serve as the basis for therapies for COVID-19. In general terms, the study aimed (a) to develop data integration methods and selection of variant characteristics to measure pathogenicity and (b) to develop data integration methods for the construction of a database of siRNAs for SARS-CoV-2. To achieve the first objective, a decision tree-based classification model was proposed to estimate the pathogenicity of variants, built through an integration process of public data of already cataloged variants with pathogenicity predictions provided by machine learning-based tools. The model was able to present a higher accuracy than the state of the art regarding the prediction of pathogenicity of variants, constituting an important tool to support health professionals, such as in the diagnosis of genetic diseases. In the second objective, data on available properties, thermodynamics, toxicity, similarity, and efficiency were combined to assemble a global catalog of siRNAs for SARS-CoV-2. The integration of diverse properties related to siRNAs in a single consolidated database is an information reference that allows the realization of simple and targeted filtering in siRNA, saving the execution of many wet-lab tests on candidate molecules for COVID-19 antiviral therapies. These studies have common features with other data integration works in aspects involving data diversity, reproducibility, and knowledge discovery. Finally, it was found that these studies have potential for clinical application, either to increase the understanding of variants related to different genetic comorbidities, in the case of the first work, or to support the development of therapies against COVID-19, in the case of second job.

The impact of extended phenotypes on the contemporary theory of evolution is controversial. The extended phenotype theory states that the expression of genes may have effects beyond the body of the individual who possesses it, affecting evolutive results of other individuals which coexist with it.The extended fitness proposes that individuals with enough genetic similarity may use the extended phenotypes of each other, thus increasing the chances of survival and reproduction of the group as a whole. This work aims to model these interactions through random scale-free networks, and investigate the impact of extended phenotypes and its effects in the reproductive success of individuals in the context of groups capable of producing and sharing them. The advantages given by the use of extended phenotypes released by similar neighbors may grant an evolutionary incentive at the group level to build and share them, and this equilibrium is measured in different simulations of behavior models.

Cryptococcosis is a fungal infection caused by yeasts of Cryptococcus spp. The infection starts when desiccated cells or spores are inhaled and reach the lungs. If the disease is not properly treated, the infection can evolve and reach the central nervous system and result in meningococcal meningitis and even death. The treatment of cryptococcosis is carried out in three stages and uses three drugs: fluconazole, amphotericin B and 5-flucytosine. Although effective, the use of these drugs can result in fungal resistance and can be toxicity for patients. This work aims to investigate the mode of action of the antifungal compound plumieridine as well as the identification of its molecular target in C. neoformans. For this, a series of in vitro and in silico experiments were carried out. Initially, a chromatographic fraction containing plumieridine was obtained from the aqueous extract from seeds of Allamanda polyantha and the presence of the compound observed through carbon and hydrogen nuclear magnetic resonance. Antifungal activity, assessed through MIC, was 0.250 mg/mL. Through virtual screening based on ligand’s similarity, chitinase was identified as plumieridine’s molecular target. Three- dimensional models of C. neoformans chitinases were created and, through molecular docking, it is observed plumieridine interacts with residues in the active site. Chitinolytic inhibitory activity assays show that activity is significantly reduced in the secreted fraction and soluble cell fraction, however, the chitinolytic activity is little reduced by the presence of plumieridine in the insoluble cell fraction, where higher concentrations of the compound are needed. Although plumieridine is able to inhibit chitinolytic activity, the compound does not appear to affect the transcriptional levels of C. neoformans chitinases: only transcription of CHI22 was reduced in the presence of plumieridine. The treatment with plumieridine still alters the distribution pattern of the chitooligomers in the cellular wall: from a polarized pattern to a diffuse pattern through the wall. The results confirm the prediction of virtual screening and show that inhibition of chitinolytic activity by plumieridine results in incomplete cell division and, consequently, cell death.

Changes in the amino acid sequence may result in alterations in the three- dimensional protein structure, which may lead to partial or complete loss of function. One way to represent the chemical interactions between all amino acids in a protein is through the construction of residue interaction networks (RINs). In RINs, a graph represents the protein 3D structure, with the nodes as amino acid residues, and the edges as the physicochemical interactions between amino acids. We hypothesize that the comparison between RINs of the same protein in different conformations can be used to evaluate the effects of mutations and polymorphisms, as well as for the analysis and validation of theoretical protein models. Therefore, the present work aimed to build a tool to compare different RINs for a protein and to use such data to estimate conformational differences between proteins and also validate models generated by homology modeling. RINs were created using the RING 2.0 (Residue Interaction Network Generator) program. The tool developed for this purpose, called Comparator of Residue Interaction Networks (CoRINs), compares all RIN nodes generated from different structure files (PDBs) of the same protein, taking into account position, chain and residue, as well as their interactions with the other amino acids. The tool also presents a plot that estimates the variation of interactions formed by each residue, which we propose as an estimate for the conformational alterations of that protein site, from a set of compared PDBs. As a possible application for this tool, we used a dataset with oncogenes and tumor suppressor genes with their respective reported mutations mapped according to the connectivity deviation of each residue. Then we retrieved the different conformations for each resulting protein from a bank of structural conformers and constructed the RINs using the software RING 2.0 and compared them with CoRINs. The results show that mutations occurring in the tested oncogenes are more likely to occur in protein sites with a more significant deviation in the mean number of chemical interactions. Additionally, most of these genes’ mutations annotated as pathogenic and associated with clinical cancer cases occurred at sites with the most significant changes in chemical and physical interactions. These results demonstrate that the CoRINs tool can be useful in identifying non- covalent interactions essential for protein structure maintenance and in evolutionary studies, such as in the maintenance of homologous proteins function with high sequence divergence, as well as for the comparison and validation of theoretical structural models.

Arapaima gigas, known as pirarucu, is considered one of the largest freshwater fish in the world, with a notable interest in the aquaculture due to its particular biological characteristics, including its rapid growth in its early years. In recent years, despite the massive availability of data from sequencing projects, few have addressed the taxon that includes this species. The present study was developed aiming characterize the transcriptome of this species, through an exploratory transcriptional analysis and patterns of gene expression related to specific gene profiles, in addition to highlighting sex-specific genes. By cDNA sequencing of 12 different tissue samples from Arapaima gigas, a reference transcriptome was assembled with a genome-guided assembly strategy. The gene expression profiles of different male and female tissues of adult specimens were analyzed. Pipelines such as Hisat2, Braker2, Trinity, Diamond and mygene were used for the assembly and annotation of genes, as well as clusterProfiler and KEGG tools for functional enrichment analysis and animalTFDB for identifying transcription factors. In this study we highlighted a set of annotated genes which may be potential candidates to biotechnological products, as they are involved in individual tissue phenotypes, sexual dimorphism processes, and in regulation of process that can explain their unique morphological characteristics. This study can also substantially conduct further analysis.

Since 1990, epidemic spread has been the subject of many studies based on sta- tistical physics methods. The dynamics of these epidemic processes, typically of non- equilibrium, consist of competition for active (infected hosts) and inactive (uninfected hopedeiro) health status. The transition between these active (epidemic) and inactive (non-epidemic) states allows the analysis of the critical point and exponents of the sys- tem (universality class). In this thesis, the critical properties of two epidemic systems are investigated: The first compound of two population species that are human with uninfec- ted hosts (H) and infected hosts (Hi) and that of vectors composed of non-infected vectors infected (V ) and infected vectors (Vi), which spread independently in a one-dimensional network, at D rates, following a dynamic probability rule, where the cure rates of vectors and individuals are respectively φ and λ. A second epidemic system, known as suscep- tible infected susceptible (SIS), in a complex network with high aggregation factor and contamination rate λ. For both models, computer simulations are used using the Monte Carlo Method to obtain the data and perform a finite-size scale analysis to estimate cri- tical properties. The conclusion of this work is the analysis of critical points and critical exponents. It is expected to define a new class of universality and a parallel with the methodology used by epidemiology to combat infectious diseases.

White-leg shrimp (Penaeus vannamei) is the most widely cultivated species in
aquaculture in the world. Commercial cultivation usually occurs at high densities, which
favors the selection of virulent pathogens, causing epidemic outbreaks. Among the
pathogens that cause shingles, the virus that causes White Spot Syndrome Virus
(WSSV) is known for outbreaks that can result in more than 80% of mortality in less
than a week. As a result, the use of preventive strategies that allow the identification and
monitoring of microbiota in crops has become increasingly necessary, especially in
intensive systems. Recently, the use of metagenomics has been suggested for
monitoring in aquaculture. Several studies have used 16S metagenomics to study the
microbiota associated with healthy or infected shrimp with specific pathogens. Other
studies have addressed the metagenomic shotgun to discover new viruses. The
metagenomic shotgun is potentially more informative than the metagenomic by marker
genes, allowing the retrieval of genomic information from the host and its symbionts,
including viruses, whose composition may act as bioindicators of the disease stage. In
this study, the shotgun metagenomic was used to analyze the caudal muscle of a P.
vannamei specimen infected by WSSV. Taxonomic and functional classifications were
made to obtain the respective profiles of the metagenomic data. P. vannamei and WSSV
were the most abundant organisms in the classification by reads. In the analysis of the
contigs, greater abundance of contigs was observed for shrimp, bacteria and WSSV,
respectively. Functional classification was performed using the MEGAN software and
resulted in few representative groups of protein functions, which were not sufficient to
establish a functional profile of the sample. A taxonomic classification from the
BLASTx was also performed with the MEGAN and presented results similar to the
classification using BLASTn. The BLASTn results enabled the assembly of the
complete mitochondrial genome of P. vannamei. This study provides support for the use
of the shotgun metagenomics as a tool for the monitoring of the microbiota in shrimp
cultures, and it is possible to simultaneously retrieve information useful for population
genetics (through the mitochondrial shrimp genome) and the monitoring of symbionts
and pathogens , such as bacteria and WSSV.

Chagas disease kills over 10,000 people per year and approximately 8 million people are infected by Trypanosoma cruzi. The reference drug for treatment of the disease, benznidazole, is the same since the 70s. In recent years, many CYP51 inhibitors were tested against this parasite’s target. One of them, posaconazole, was even tested in clinical trials that unfortunately were not successful. Nevertheless, there are still many evidences that CYP51 is a great potential target to treat T. cruzi infection.  The research for new effective molecules that can cure the chronic phase of the disease is essential. 2D and 3D-Quantitative Structure Activity Relationship (QSAR) studies were conducted in this work to create three QSAR models using the chemical structures of 197 published compounds that already went through either in vivo or in vitro tests. After the analysis of the models, new analogues not yet synthesized were suggested here and had their biological activity and synthetic availability assessed. 

Alport syndrome (AS) is a genetically rare, heterogeneous and hereditary pathology associated with germline mutations in collagen type IV genes (COL4A3, COL4A4 and COL4A5). Characterized by progressive loss of renal function, hearing and eye damage during early childhood, the progression of the disease progresses to a terminal renal disease often associated with renal failure. Studies aimed at early diagnosing individuals with this nephropathy may lead to appropriate treatment and thus improve life expectancy. Efforts are currently underway, focused on the genome of patients, to create diagnostic tests for rare diseases/syndromes. From this perspective, mutations, genes and metabolic pathways involved with the pathology is crucial to understanding the complexity of these diseases. Thinking about corroborating the findings and studies about AS, the exome sequencing of two families from Rio Grande do Norte (RN), both composed of 4 individuals, was performed. Through the GATK and VARSCAN2 software, variants were called followed by a screening of deleterious variants identified by an in house script. The results pointed to two deleterious variants in the genes that form the type IV collagen α3 and α4 chains (a stop codon in COL4A3 and frameshift in COL4A4) leading to premature protein truncation. Both variants were detected in homozygous state in the probands and heterozygous in the other family members. Additionally, a broad region of runs of homozigosity (ROH) involving the COL4A3 and COL4A4 genes was detected in both probands of both families. According to the findings of deleterious variants in the COL4A3 and COL4A4 genes in ROH regions, these variants are now related to SA so that similar observations can serve as support for possible targets in the creation of new diagnostic tests and for the service of Genetic Counseling.

The consequences of lead poisoning are diverse and relevant to human health. Reaching all organ systems, it mainly afects the nervous system, with severe and irreversible implicatons of neurodevelopment, memory consolidaton, and learning processes in children. They interact with cellular components in many ways, afectng ion-binding proteins, transducton signaling proteins, transmembrane ion channels, and transcripton factors. If in one hand, the symptoms of lead poisoning are well known, on the other hand, we have a lack of the systemic efects and its impact on neuronal cell transcripton modulaton. In order to investgate such efects from a systems biology perspectve, we applied the transcriptogramer R/Bioconductor package pipeline to evaluate the transcriptonal profle of lead acetate- treated human neural progenitor cells (NPCs) 30μM for 26 days. The transcriptogramer algorithm is designed to identfy functonally associated and diferentally expressed gene groups in case-control experiments in an unsupervised way. It was able to identfy eleven diferentally expressed clusters between days 3 and 11 of the lead treatment. Of these, seven presented negatve regulaton of several cellular systems involved in cell diferentaton, such as cytoskeleton organizaton, RNA and protein biosynthesis, characterized by large and tghtly connected networks. The four clusters that were positvely regulated presented sparse and poorly connected nodes, mainly related to transcripton, transmembrane transport, and signal transducton. In the subsequent period, involving days 12 to 26 of treatment, it was possible to observe a massive alteraton of the cellular transcripton profle with interference in all layers of gene expression regulaton. Thus, our results suggest that lead induces signifcant transcriptonal modifcatons in NPCs which can be correlated to damage and/or adaptatons of various systems, all resultng from intoxicaton by this heavy metal, thus influencing the result of ES-NP cell diferentaton.

Cancer/testis (CT) genes are excellent candidates for cancer immunotherapies because of their restrict expression in normal tissues and the capacity to elicit an immune response when expressed in tumor cells. In this study, we provide a genome-wide screen for CT genes with the identification of 745 putative CT genes. Comparison with a set of known CT genes shows that 201 new CT genes were identified. Integration of gene expression and clinical data led us to identify dozens of CT genes associated with either good or poor prognosis. For the CT genes related to good prognosis, we show that there is a direct relationship between CT gene expression and a signal for CD8+ cells infiltration for some tumor types, especially melanoma. In addition, we contextualized bioinformatics in a big data scenario.

Currently there are several tools proposed for analysis of Transcription Factors (TF), such  as  TFCheckpoint,  JASPAR,  SSTAR,  GTRD,  Enrichr. However  none  of  these tools offers a complete experience in which the reliability of TF can be evaluated, that is,  if  in  fact  an  analyzed  protein  is  a  TF  and  its  association  with  the  target  gene. Numerous databases were built over time, all of them with very rich information, but the  intrinsic  complexity  of  the  data,  the  volume  of information,  problems  of  gene nomenclature  and  several  other  factors  meant  that  such  tools  did  not  offer  a complete spectrum of analysis . On the other hand,  to work with a large volume of data  requires  advanced  computer  skills.  However,  the  general  public  interested  in analyzing this data are professionals from the biological areas. Configuring itself as a barrier,  since  the  academic  formation  of  this  area  does  not  offer  in  its  curricular components  programming  disciplines.  Faced  with  this situation,  this  work  aims  to create  a  web  tool  exclusively  for  the  analysis  of TFs.  Containing  the  integration of different databases and a set of scripts to manipulate this information, along with the crucial parameters defined by the user in its analysis, Transcription Factor Analysis Tools (TFAT) was designed and developed. The core of this tool is the analysis to identify  the  key  TFs  in  the  modularization  of  gene  transcription,  that  is,  the enrichment of the regulatory TFs of a list of genessubmitted by the user, that through the  scripts  that  integrate  the  same,  consult  its  database,  identify  the  TFs  that  are associated  with  the  listed  genes  and  calculate  the  enrichment  p-value.  In  addition, the tool verifies TF reliability, makes available predictions, and converts items from a list to the Entrez Gene's GeneID or Symbol. Anotherfeature of this work is the use of TF reliability applied throughout the tool. This degree of reliability takes into account evidence from different databases, experiments, predictions and other characteristics of TFs. With a standard mode and a user-defined mode, this reliability feature allows for a full customization through filters in the queries and analysis control for the end user.

The creation of gene expression encyclopedias possibilities the understanding of gene groups that are co-expressed in different tissues and comprehend gene clusters according to their functions and origin. Due to the huge amount of data generated in large-scale transcriptomics projects, an intense demand to use techniques provided by artificial intelligence became widely used in bioinformatics. Unsupervised learning is the machine learning task that analyzes the data provided and tries to determine if some objects can be grouped in some way, forming clusters. We developed an online tool called CORAZON (Correlation Analyses Zipper Online), which implements three unsupervised machine learning algorithms (mean shift, k-means and hierarchical) to cluster gene expression datasets, six normalization methodologies (Fragments Per Kilobase Million (FPKM), Transcripts Per Million (TPM), Counts per million (CPM), base-2 log, normalization by the sum of the instance's values and normalization by the highest attribute value for each instance), and a strategy to observe the attributes influence, all in a friendly environment. The algorithms performances were evaluated through five models commonly used to validate clustering methodologies, each one composed by fifty randomly generated datasets. The algorithms presented accuracies ranging between 92-100%. Next, we applied our tool to cluster tissues, obtain gene’s evolutionarily knowledgement and functional insights, based on the Gene Ontology enrichment, and connect with transcription factors. To select the best number of clusters for k-means and hierarchical algorithms we used Bayesian information criterion (BIC), followed by the derivative of the discrete function and Silhouette. In the hierarchical, we adopted the Ward’s method. In total, we analyzed three databases (Uhlen, Encode and Fantom) and in relation to tissues we can observe groups related to glands, cardiac tissues, muscular tissues, tissues related to the reproductive system and in all three groups are observed with a single tissue, such as testis, brain and bone-narrow. In relation to the genes clusters, we obtained several clusters that have specificities in their functions: detection of stimulus involved in sensory perception, reproduction, synaptic signaling, nervous system, immunological system, system development, and metabolics. We also observed that clusters with more than 80% of noncodings, more than 40% of their coding genes are recents appearing in mammalian class and the minority are from eukaryota class. Otherwise, clusters with more than 90% of coding genes, have more than 40% of them appeared in eukaryota and the minority from mammalian. These results illustrate the potential of the methods in CORAZON tool, which can help in the large quantities analysis of genomic data, possibiliting the potential associations analyzes between noncoding RNAs and the biological processes of clustered together coding genes, as well as the possibility of evolutionary history study. CORAZON is freely available at http://biodados.icb.ufmg.br/corazon or http://corazon.integrativebioinformatics.me.

Variations in the level of gene expression are among the main causes of phenotypic diversity in organisms, including the development of pathologies and response to drugs in humans. Non-coding RNAs (ncRNAs) play an important role in the complex mechanism of regulatory networks. Although not yet fully understood, two representatives of the ncRNAs emerge in recent researches as protagonists in the development of clinical conditions. They are the microRNAs (miRNAs) and the long intergenic non-coding RNAs (lincRNAs). Thus, the present work integrated public data to catalog the vast landscape of the regulatory effects of miRNAs and lincRNAs in the human genome. Through expression Quantitative Trait Loci (eQTL) analysis, variations that had a putative effect on gene expression were identified. Association networks were also created relating the eQTL analysis results to traits of clinical and/or pharmacological relevance. Through this, associations that may continue to arouse the interest of new studies involving the theme were revealed. Mental and coronary disorders, in addition to cancer, were the most evidenced traits in the study results.

Breast cancer and a hormone-dependent disease, which has several different subtypes, patterns of gene expression and distinct manifestations (CHENG et al., 2002). According to the National Cancer Institute (INCA), in the year 2013, as deaths caused by the disease of 14,388, being 181 men and 14,207. The estimate for 2015 is 57,120 of new cases. Most breast cancers are ER + (estrogen receptor positive), ie, 17β-estradiol dependent. In this type of breast neoplasm, the number of ERα (estrogen receptor alpha subtype) is higher than  the number of ERβ (estrogen receptor beta subtype), evidencing the importance of the alpha subtype in this disease. The purpose of this work is to measure the individual binding  energies  of  ERα  residues  with  17β-estradiol  and  Diethylstilbestrol,  using  a computational simulation. For this purpose, it is employed as Doria of Functional Theory (DFT) and Molecular Fractionation Method with Conjugated Caps (MFCC). The results obtained with this work may help to characterize the interaction between the 17β-estradiol agonists and Diethylstilbestrol with ERα. The results obtained showed the residues with the most significant energy values are: GLU353, LEU391, MET343, LEU346, MET388, ARG394,  PHE404,  HIS524,  ASP411,  LEU525,  ARG352  and  ARG548. These  results help characterize, through the information obtained, an interaction between 17β-estradiol and Diethylstilbestrol with ERα and, in turn, can be used as a basis for studies, structural drug design, modulate existing drugs, such as for the design of new drugs.

Ewing Sarcoma (ES) is a rare malignant bone tumor with high propensity to metastasize occurring most frequently in adolescents and young adults. There is no ES cell of origin identified só far and the hallmark of this cancer is the occurrence of a chromosomal translocation between the chromosomes 11 and 22 that results in an aberrant transcription factor through the fusion of a gene from FET family and ETS family, commonly EWSR1 and FLI1. The translocation is associated with chromatin alteration, leading to a significant disturbance in the cell transcriptome. The regulatory mechanisms behind the observed ES transcriptional alterations remain poorly understood. Here, we inferred the transcriptional regulatory network of Ewing Sarcoma and identified 7 transcription factors as potential master regulators. According to our results, these 7 master regulators are organized in two clusters: one composed by PAX7 and RUNX3 and other composed by ARNT2, CREB3L1, GLI3, MEF2C, and PBX3. The master regulators inside each cluster are agonists among each other andboth clusters show antagonism between them. Based on transcriptional data, we classified ES patients of two cohorts according to the activity of each of the seven regulons. High regulatory activity of PAX7 and RUNX3 is associated with better overall survival and high regulatory activity of ARNT2, CREB3L1, GLI3, and PBX3 is associated with worse overall survival. This work contributes to a better understanding of the regulome of Ewing Sarcoma, indicating putative master regulators that can lead to potential prognosis prediction and key factors of tumorigenesis.

In the last decades, advances in whole genomic approaches lead to the identification of a vast number of cancer-related mutations. High-throughput estimations of the impacts of cancer mutations in the protein structure are not an easy accomplishment, and most studies are limited to one-by-one whole structural analyzes. Moreover, there are still many challenges on the way to the precise and automated prediction of pathogenic mutations. Therefore, understanding the structural impact of a particular amino acid change is of great importance for cancer medical research. However, most studies have been emphasizing sequences and structural modifications based on chemical characteristics of amino acids and not fold features, in which the conservation of non-covalent interactions play a significant role. Henceforth, in the present study, we used residue interaction networks (RINs) for large-scale analysis of cancer missense mutations in order to infer their effects on the conservation of non-covalent interactions. We hypothesize that changes in highly connected amino acids are more likely to cause deleterious mutations. To evaluate this, we retrieved cancer missense mutations from COSMIC (cancer.sanger.ac.uk/cosmic) and TCGA (cancergenome.nih.gov) databases and mapped them to their respective structures retrieved from Protein Data Bank (rcsb.org). Then, RINs were constructed from the obtained pdb files, and network parameters such as the node's degree, edges' type, clustering coefficient, betweenness weighted were assessed and plotted using R scripts. Later, we compared these results against reported missense single nucleotide polymorphisms retrieved from dbSNP (www.ncbi.nlm.nih.gov/projects/SNP/) and to pathogenic and non-pathogenic cancer mutations from ClinVar (www.ncbi.nlm.nih.gov/clinvar/) databases. Our results demonstrate that the distribution of mutations per degree (node connectivity) varies significantly compared to random Monte Carlo simulations and also to the distribution of a set of human single nucleotide polymorphisms (SNPs), tending to remain at nodes with lower connectivity. Besides, the proportion of deleterious mutations was significantly increased in nodes with a high degree of connectivity when two different criteria were used for their classification: proportions of software predictors (Ndamage) and clinical classification obtained from ClinVar. Taking into account these results, we can conclude that the changes in the highly connected amino acids are indeed more likely to generate deleterious mutations, due their higher proportion of occurrence in these nodes. Our results also indicate that the conservation of non-covalent interactions is an important parameter to consider in assessing mutations effects and RINs analyses can be used as an additional parameter to aid in the prediction of deleterious mutations in cancer.

Big Data is a term used to characterize the growing volume of existing data on different topics, whether they are biomedical or not. The enormous volume of biological and biomedical data generated daily, one of the main barriers will be an analysis of these data. The development and use of computational tools that allow the analysis of data through techniques such as Text Mining. Text Mining, a Data Mining strand, can be defined as a method that allows the extraction of relevant information contained in text. In order to allow a differentiated analysis of the data, whether these clinical data or not, a simple algorithm was developed, which allows the analysis of this data without the need of correlation with existing databases, nor the creation of new databases. From this algorithm, a WEB tool was developed so that anyone can access the algorithm (even without the knowledge of computational techniques) and promote the analysis of their data. The Integrate Paired Tool (IPT) algorithm was written in R programming language and uses Data Mining and Text Mining techniques for analyzing clinical data, not restricting its analyzes only to these specific data. IPT promotes pairing of terms by analyzing the existing frequency between data pairs, from a user-supplied .csv file. In addition, the WEB tool was developed from the languages JavaScript, HTML5, CSS and PHP. The algorithm reads the .csv file and pass through it by pairing its terms two by two, regardless of whether the columns are different sizes or incomplete until all columns are paired. After all the groupings, a value is assigned to each grouped pair, adding all pairs with the same frequencies and generating another .csv file containing the existing interactions and their respective frequencies. After the relations and their appearance frequencies are formed, a graph of interactions (in R) is shown on the WEB tool screen, so the user can do their analyzes, in addition to the .csv file with all interactions and frequencies. This graph and this table can contain variable information, depending on the percentage that the user chooses in the IPT tool. This .csv file with interaction and frequency data can be used by the user in other network visualization tools, such as Gephi, for example. For the purposes of tool testing, a data from a neonatal was used. The IPT proved to work well and reached the objectives of the research, and as future goals, we will have the hosting of the tool in the page of the Program of Postgraduate in Bioformtics of UFRN, the analysis of other data and a possible integration of the pre-processing of the data within the IPT itself.

Lead is an important heavy metal used worldwide in several applications, especially in industry. People exposed to lead can develop a wide range of symptoms associated with lead poisoning. Many effects of lead poisoningwere reported in the literature, showing a compromising of whole body health, with symptoms related to cardiovascular, immune, bone, reproductive, hematological, renal, gastrointestinal, and nervous system. However, the molecular lead targets as wellas the pathways affected by lead poisoning are not completely described. The aim of this study was to construct a map of metabolic pathways impaired in lead poisoning byevaluating which biomolecules are directly affected by lead. Through manual literature curation, we identified proteins which physically interact with lead and subsequently determined the metabolic pathways those proteins are involved with. At total, weidentified 23 proteins involved with heme synthesis, calcium metabolism, neurotransmission, among other biological systems, which helps to understand the wide range of lead poisoning symptoms.