The box plot presents the distribution of correct reads in the verification NGS run of retrieved bead. A total of 96,484 molecules (1,010 kinds) were arranged with respect to the quality score and depth. Blue boxes show the correct read distribution considering both indel and substitution mismatches, whereas red boxes take only substitution into account. The median percentage values of the blue boxes in group 1 were in the mid-eighties, which is almost 10% less than those of the red boxes in the same group. However, these values gradually increased in the group of higher quality score reads, reaching the almost constant red box values for group 3 (95.7%). These results indicate that the majority of indel mismatches are sequencing errors that originated from error-prone homopolymer pyrosequencing. Therefore, we can say that the percentage of correct DNAs retrieved is >96%.