Comparison of topology-elimination algorithm and standard ranked list. A performance comparison between standard GO list from DAVID and Ontologizer output with the topology-elimination algorithm. We show the difference in the cumulative sum of positive GO classes across the top ranks. X-axis shows the rank, whereas the Y-axis shows the difference. Positive value indicate better performance by the Ontologizer algorithm and negative value better performance by the DAVID GO list. We summarize the comparison with 20 datasets showing five percentiles for each rank position: median (black line), 25 and 75 percentile (blue line) and minimum and maximum (cyan line). Notice that topology-elimination shows better performance across the top-15 ranks.