BDBComp
Parceria:
SBC
A semi-automatic methodology for localization of short mitochondrial genes in long sequences

Rafael SantosJosé Humberto Machado TamborLuciana Campos PaulinoAna L. C. Bazzan

Identification of short genes in long sequences using similarity measures (e­values and scores in BLAST queries) can be difficult in mitochondrial genomes since the similarity results of some genes can be shadowed by neighbor matches with higher similarity values. The same could happen for genes with relatively low similarity but which can be considered of interest in a particular study. In order to locate and identify those genes, a manual analysis of the similarity search results must be done, which can be time-consuming and error-prone. In this report we present a methodology which aids researchers on the location of those genes by semi-automatically masking subsequences corresponding to genes that were already identificated and limiting subsequent searches to the regions that did not present any result in previous steps. A tool that implements this methodology was created and used in some database searches using a sequence obtained from a mitochondrial genome. We expected that analysis using this tool would be easier if not faster than the manual analysis. Some results of the use of the tool are presented and compared with results obtained by manual similarity searching of BLAST results. As expected, the proposed tool didn't present new results (i.e. different from the ones found in the manual analysis), since both rely on the same search mechanism, input and parameters, but the results were clearer in the sense of not being cluttered with similar results, and the shorter genes could be located more easily on the final similarity report. Some comments on the classification of this tool as a software agent are also shown. Suggestions for improvements of the methodology and tool will also be presented.

http://www.lbd.dcc.ufmg.br:8080/colecoes/wob/2002/010.pdf

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD