Scatter plot of MEGABLAST search times with and without partial retrieval. 163 human ESTs from UniGene cluster 235935 were searched against all human chromosomes . On the x-axis are times for the baseline application; on the y-axis are times for the new blastn application. Sequences with the best improvement are those furthest to the right, and they also matched the largest number of subject sequences. A word size of 24 was used for the runs as well as database masking with RepeatMasker. Three searches were done with both the baseline and blastn application for each data point, and the lowest time for each application was used.