Shown are the sizes of the left arm (L-arm), the right arm (R-arm), the number of independent experiments, and the number of “correction” events from the total number of G418-resistant clones for each targeting vector. The 95% confidence intervals for correction events are shown. All of the vectors displayed HR frequencies that were statistically indistinguishable from the parental “knock-in” vector (4 correctly targeted events of 220 G418-resistant clones. For this calculation [*], we divided the parental targeting efficiency by 2, because we can only detect HR events that occur on the mutant allele with the “correcting” vectors). We have transfected the same RW-4 ES cell used here with targeting constructs representing 255 independent loci in our ES core. The design of all vectors was standardized to include at least 2-kb isogenic targeting sequences in each arm, with selection provided by the same PGK-neo cassette used in this study. A total of 39 085 G418-resistant clones have been evaluated; 606 of these (1.55%) have been confirmed to contain correctly targeted alleles, an HR efficiency that is statistically the same as the efficiencies measured in this study (E. Ross, J. Mudd, D. George, and T. J. Ley; http://medicine.wustl.edu/~escore/htmldocs/gencon.htm#targrec).

One correctly targeted clone also harbored additional randomly integrated vector sequences elsewhere in the genome (data not shown).