When plotting the distribution of gap sizes in assemblies of several genomes I found out that the genomes that were assembled by ALLPATH-LG have the special pattern of distribution of gap sizes. For example gaps whose sizes are divisible by 5 occur significantly more often. In the assemblies performed by the other software this pattern was not revealed. I plotted a range of gaps of sizes less than 200 but the discussed pattern holds on all gap sizes. The first to figures illustrate the pattern, the figures 3 and 4 show gap pattern produced by other assemblers. Does anybody know if this specific gaps size distribution can be explained with incorrect parameters used for ALLPATH-LG run, or is it a known issue?