The top percentiles of single organisms ranked lists obtained with orthologous probes are strongly enriched of orthologous sequences. 100 orthologous (CLOE) and 100 randomly chosen (Random) EST pairs were used to rank the ESTs in the human and mouse datasets on the basis of expression similarity. The ranked lists were divided in 1% rank intervals, and the average number of human ESTs in a given rank interval with at least one orthologous EST in the corresponding mouse rank interval was determined. The average number of these ESTs in the first top 10 rank intervals was plotted. Error bars = standard error.