RESUMO
MOTIVATION: To identify genetic conservation relative to precise aspects of developmental diversity, an essential question in computational biology, we developed a new comparative method that allows conserved modules for the best balance between protein sequence similarity and gene co-expression to be constructed, in invertebrates. RESULTS: Our method, referred to as the best-balance constraint procedure (BBCP), yielded 719 functionally conserved modules (FCMs) comprising 2-23 gene pairs. These modules were consistent with the developmental roles of orthologues as inferred from Gene Ontology, RNAi knockouts, InterPro and process-specific microarray data. New relationships were defined between genetic conservation and developmental diversity. Novel gene associations were indeed found in 94% of the FCMs, 150 modules being completely new. A significant proportion of the FCMs (18%, 132 modules) described cell type-specific mechanisms, comprising neuronal, muscle and germ cell signaling, new associations being found in 125 modules. Also found were gene associations for cell fate specification activities previously not highlighted by computational means, e.g. in FCMs containing homeogenes. These data indicate that highly discriminative description of genetic conservation can be deduced using BBCP, and reveal new correlations between cellular and developmental diversity and gene essentiality in invertebrates. CONTACT: christian.neri@broca.inserm.fr SUPPLEMENTARY INFORMATION: For supplementary information, please refer to Bioinformatics online.
Assuntos
Proteínas de Caenorhabditis elegans/genética , Proteínas de Caenorhabditis elegans/metabolismo , Proteínas de Drosophila/genética , Proteínas de Drosophila/metabolismo , Perfilação da Expressão Gênica/métodos , Regulação da Expressão Gênica no Desenvolvimento/genética , Animais , Proteínas de Caenorhabditis elegans/química , Sequência Conservada , Proteínas de Drosophila/química , Evolução Molecular , Invertebrados , Homologia de Sequência de AminoácidosRESUMO
We present a fast algorithm for sequence clustering and searching which works with large sequence databases. It uses a strictly defined similarity measure. The algorithm is faster than conventional EST clustering approaches because its complexity is directly related to the number of subwords shared by the sequences. Furthermore, the algorithm also works with proteic sequences and large sequences like entire chromosomes. We present a theoretical study of our approach and provide experimental results.