Discovering a gene can be extremely difficult if its sequence differs significantly from known genes. After all, most approaches depend on the assumption that the primary sequence of one gene strongly resembles that of another, known gene — what if it doesn't? On p. 1107, Junhyong Kim and John Carlson discuss a novel in silico approach — e-Genetics —that can search for related proteins on the basis of structural similarity. They have used e-Genetics to identify Drosophila odor and taste receptor families, which other strategies had failed to uncover. Both families are G-protein-coupled receptors (GPCRs); the authors therefore developed a way of mapping an n-dimensional protein space in which GPCRs occupy a particular region, using parameters such as the periodicity of hydrophobicity/polarity along the length of the protein. In tests on SWISS-PROT sequences, the algorithm correctly identified 96% of GPCRs and did not misclassify a single non-GPCR. Kim and...

You do not currently have access to this content.