Using ancestral sequences to uncover potential gene homologues.

Gene homologues between distantly related species can be difficult to identify. We test the idea that inferred ancestral sequences could aid in finding gene homologues. Ancestral sequences are inferred by aligning gene homologues on a known tree and estimating the most likely amino acid for each position at each node in that tree. BLAST(R) and HMMER are used separately and together with ancestral sequences to search the genome sequence databases of Encephalitozoon cuniculi, Entamoeba histolytica and Giardia lamblia for RNase P protein homologues. RNase P proteins (Pop4, Pop1, Pop5 and Rpp21) have been reported in humans and at least two other eukaryotic species but have yet to be identified in the above genomes. Using ancestral sequences reconstruction (ASR) for these proteins, we successfully identified putative homologues from E. cuniculi, Ent. histolytica and G. lamblia. In some cases, the use of ASR outperformed BLAST and HMMER. Overall, including ancestral sequences in searches with BLAST and/or HMMER was the most successful approach in the recovery of potential RNase P protein gene homologues, making this a useful technique in early homologue identification.