Duplicate sets of blood samples were provided from the original fibroblast donor dog (Tai, an Afghan hound), the surrogate mother (a Labrador retriever) and Snuppy, the putative clone. Samples were drawn in heparinized tubes and delivered to us on ice overnight. Collection and mailing of samples was supervised by In Kwon Chung, a member of the investigative committee at Seoul National University in South Korea. Samples were not provided from the oocyte donor, which was unavailable for sampling. Samples were coded by a third party and laboratory personnel were blind to sample identifiers.

In addition to these six samples, the test panel included previously purified DNA samples from 11 Afghan hounds collected in the United States and registered with the American Kennel Club (AKC); eight of these shared no common parents or grandparents, and the other three were half-siblings that shared a common sire. Pedigree analysis revealed that one of the American-collected dogs was a first cousin of Tai, the fibroblast donor, and that five of the others had distant maternal and paternal relatives in common with him. Other samples on the test panel included a pure-bred female Labrador retriever, purportedly unrelated to the surrogate mother, and Tasha, the boxer dog used for the reference canine sequence4.

We tested both nuclear and mitochondrial markers. Nuclear markers included 16 microsatellite markers routinely used for canine paternity testing by the AKC5,6. (A seventeenth marker was discarded because it failed to amplify from 25% of the DNA samples.) For all nuclear markers tested, Snuppy and Tai, the clone and donor, had identical genotypes (Table 1).

Table 1 Microsatellite genotypes of 16 dogs, including the putative clone, the nuclear-DNA donor and the surrogate mother

The probability that the putative clone should have precisely the same genotype as the donor was computed for different assumptions regarding the relatedness of the sample and the donor7 (Table 2). In all cases, the allele frequencies were computed from a sample of 11 AKC-registered Afghan hounds plus the donor. According to genetic maps of the canine chromosomes, the 14 mapped markers were unlinked to one another, so each microsatellite was treated as an independent locus8,9.

Table 2 Probability of exact allelic matching at 16 microsatellite loci in a population of Afghan hounds

The match probabilities ranged from 7×10−14 for unrelated dogs to 4×10−4 for those with a specific inbred pedigree. A higher degree of inbreeding would increase the match probability further, but the donor does not seem to be extremely inbred; both the donor and Snuppy are heterozygous at 8 of the 16 markers, which is not significantly different from the number of heterozygous markers expected, given the observed allele frequencies and no inbreeding.

Mitochondrial D-loop sequencing revealed 26 variable bases within the 614 analysed (Table 3). Snuppy and the donor dog differed at 12 of the 26 sites. Nine of the Afghan hound sequences disagreed with each other at only one base (position 548) and differed from the donor by only three to four bases. Also, the two Labrador retrievers had identical mitochondrial sequences that differed from the donor by only three bases. The sequence from Snuppy differed from that of any other dog at 9–14 sites.

Table 3 Mitochondrial genotypes at 26 polymorphic loci within the canine mitochondrial D-loop

These data are consistent with Snuppy being a genetic clone of the donor dog Tai. Our analysis rules out most feasible alternatives to a true clone, such as the production of a delayed twin, which would have produced dogs with the same mitochondrial D-loop sequence, or an animal resulting from extreme inbreeding, which would have yielded dogs that were homozygous at more than the observed eight loci.

Conclusions drawn from these results are subject to caveats. First, we did not witness the drawing of the blood samples, which was done under the supervision of a third party. However, no obvious hypothetical manipulation of the samples would have generated the results described here — perfect matching of the nuclear markers, and distinct differences between the mitochondrial sequences for the donor and Snuppy. Second, we were not provided with samples from the oocyte donor, although tissue samples from this dog have been tested by investigators at Seoul National University10. Without this sample, we are unable to confirm the original experimental details1, or to say with certainty that the mitochondrial variants observed were those that were expected.

Finally, our statistical analysis is based on a limited number of Afghan hounds. A larger number of unrelated individuals might have provided a more precise estimate of population allele frequencies. However, given that the dogs tested are representative of this relatively restricted breed and that many share a partial heritage with the donor, the statistical conclusions are conservative.