Desenvolvimento de ferramentas para a identificação de marcadores moleculares e imunológicos a partir de dados genômicos como alvo para o diagnóstico de doenças parasitárias

Abstract

Infectious diseases caused by protozoan parasites is a major public health problem, especially in poor or developing, causing millions of deaths annually. For effective monitoring and control of these diseases it is essential to develop precise diagnostic methods. The identification of immunological and molecular markers, such as microsatellites and epitopes, allows rapid selection of potential targets that can be used in diagnostics, vaccination protocols and immunotherapeutic. Experimental methods for the identification of immunological and molecular markers have high costs and require long periods of experimentation. Given the large amount of genomic and protein sequences available in public databases, in silico methods for identifying these markers have been employed as an alternative approach. Recently, many methods and tools to identify immunological and molecular markers have been developed. Epitope prediction tools such as PREDITOP, PEOPLE, BEPITOPE, BepiPred, ABCpred, BCPred, BayesB and BEST were developed using machine learning techniques. These tools have better results in comparison to tools that do not use this approach. However, they are not accurate enough when applied to protozoa data, mainly due to the small protozoan datasets used for training. To identify tandem repeat markers, there are several available tools, such as IMEX, MISA, Mreps, SciRoKo, Sputnik and TROLL. However, besides identifying repetitive regions, other features have to be displayed to assist researchers in the recognition and analysis of molecular markers. In this work, we have developed tools for the identification of molecular and immunological markers from genomic data in order to seek new targets for the diagnosis of parasitic diseases. In the first part of this work, it presents a web and stand-alone tool, entitled ProGeRF (Proteome and Genome Repeat Finder), and developed to identify tandem repeats as molecular markers. The second part of this work aimed to verify whether the performance of in silico prediction tools of linear B-cell epitope could be impacted by using distinct training datasets (bacteria, viruses and protozoa). The ProGeRF tool is an efficient, fast, accurate, easy to use, either in stand-alone or web tool, provides a graphic display and allows filtering the results. Besides, it is able to run in large genomic and proteomic data. When compared with MISA, TROLL, TRF, Sputnik, SciRoKo and GMATo, ProGeRF can identify a larger number of repetitive elements and is faster than the majority of the other tools. Regarding the prediction of linear B-cell epitope from protozoa data, tools trained with bacteria data generally leads to random predictions. However, when trained only with virus data, some species of protozoa showed a significant degree of efficiency during B-cell epitope prediction. The best results were obtained when the training dataset contained balanced data from the three taxon or data only from protozoa, highlighting the importance of optimizing the performance of tools for B-cell linear epitope using adequate training datasets.