Database searches are very difficult to interpret-and I find homology (similarity) figures in papers frequently border on the unbelievable, but give no indicators of confidence, even statistical. In response to the post by brett, the P-values are an indicator of the likelihood that the score of the match could differ from an expected score. Low P-values indicate clear similarity, high p-values are dubious. I find that some matches of real biological significance have p-values as low as 0.3 (BLAST searc
es). Lower values might be signifcant, but I doubt it. Some higher values are probably not significant.