finding highly divergent orthologs using exonerate

Question

[Emily]

Hi I would like to find a protein ortholog that is highly divergent from drosophila to human, that using tblastn with Drosophila ortholog to blast cannot even identify the human orthologs unless using a very low e value (0.5) and word size of 2. it could be due to the fact that it is much shorter than the human ortholog When using the p2g of exonerate with drosophila ortholog as query, I could not identify any sequence from human. I've tried reducing the identity percentage to 10, reducing proteinwordthreshold and proteinhspthreshold to 2 and still could not identify any aligned orthologs. Are there any approach I could increase the sensitivity of exonerate p2g especially for highly divergent proteins?

[dthorbur]

It might be worth trying to cluster sequences with a few intermediate species too. Then the centroid of the cluster can theoretically be one of the evolutionary intermediates rather than the edges of the cluster. That said, it's such a large evolutionary gap between drosophila and humans this still might not be possible.