Anyone looking to experimentally tie a disease or a trait to a gene knows that the stars are fading for the DNA features called "microsatellites", while enthusiasm for the markers called SNPs ("snips") is building. The thing is, with all that scientific glory to be had, who has the time for comparison shopping? Gearing up to trace the chromosomal roots of a complex and perhaps only hypothetically genetic condition like "susceptibility to psychological depression" is sort of like readying to cross Antarctica . Since Amundsen et al. we've been going in by sled dog, but those snowmobiles are looking pretty darn attractive.

As many teams forge into the genetic unknown without microsatellites to guide them, Giulia Kennedy and colleagues have been at the test track putting SNPs and microsatellites through a side-by-side comparison. Word from the tests is good for our friends in the field.

Kennedy is Director of Genomics Collaborations at Affymetrix and a scientist at the Santa Clara HQ who co-designed the company's SNP arrays, including the Mapping 10K Array and more recently the 100K Set. When Kennedy learned that a team led by Sally John at the University of Manchester had sampled 550 individuals from 157 families for a microsatellite linkage analysis of rhuematioid arthritis, she was eager to run a SNP analysis alongside it. Theorists told her she had more than enough SNPs on the 10K to outpace microsatellites, but Kennedy likes to see the checkered flag and smell the burning rubber. "No one had actually done the proof-of-principle experiment," Kennedy says. "We wanted to do it head-to-head and in the same cohort."

Kennedy liked John's targeting of rheumatoid arthritis (RA) because the genetics of it are far from obvious and suggest the involvement of disparate loci. Another challenging aspect is RA's late-life onset, which limits the help that family-tree relationships traditionally provide in establishing genetic links and mapping genes. Because of the late onset, RA sufferers rarely have living parents from whom investigators can collect blood or DNA. Additionally, it's typically too early to say whether the sufferers' children will themselves show the disease or not. The nearly tree-free environment of RA makes its analysis complicated, like complex traits and conditions in general , Kennedy points out. In addition, because a locus on chromosome 6 already had been nailed down, RA lends a prospective investigation a built-in control.

John was happy to have Kennedy pull her SNPs up into the neighboring lane. In spite of the clarity with which microsatellites had implicated the one mapped locus, other loci stood in limbo, because various analyses that had already been reported were in conflict about the loci's relevance and position. John was investigating a larger cohort of more interrelated individuals than previous teams had examined, but Kennedy's parallel analysis against an alternate set of landmarks would add welcome weight to any results.

As in other whole-genome studies that take microsatellites as their points of reference, John's team was sequencing each subjects' DNA at 350 positions across the 23 chromosomes—the loci of the variable repeats called microsatellites. In the new cohort the locus on chromosome 6 that had previously been implicated in RA again stood out starkly, but so too did other loci that had emerged from just a subset of the prior studies. Sequencing, calling the genotypes and running the analysis to link particular loci with RA, took the team eighteen months.

As Kennedy recalls John's team had a head start, but Team SNP still lapped them. "We were able to blow through that study very quickly," she says. Of course, there was no rule preventing Team SNP from using Affymetrix technology to sequence all 11,245 SNPs of each study participant with a single chip and call the genotypes automatically by computer. But with the results in hand they linked the same loci that John's team had, in addition to four more. Among the extra is a locus on the X chromosome—especially intriguing in light of the higher incidence of RA in women ("XX") versus men ("XY").

Now John has members of her team making a bee-line for the X locus, which rose to prominence for the first time against the SNP terrain. Future cloners of the previously spotted chromosome-6 locus, meanwhile, have less ground to cover as a result of Team SNP's 10K mapping. While microsatellites had mapped the locus to within 50 centimorgans, Kennedy and colleagues found SNPs within 31. What's more, the results of their analysis were essentially the same whether or not they used a knowledge of certain subject's relation to one another to screen out calls that looked unlikely.

While Kennedy was pleased with the performance of the SNP analysis , she was equally pleased with how small an amount of fuel she had needed to run it: Only 250 nanograms of DNA from each participant—equivalent to the amount extractable from a couple drops of blood. Mileage like that, as Kennedy points out, can eliminate the need to create immortalized cell lines, or conversely the need to relocate subjects when a locus proves hard to genotype and not enough sample remains for resequencing.

This efficiency was partly a trick of sample preparation, Kennedy states. She and her team had circumvented most of the work that usually goes toward repackaging an individual's 6.6 billion base pairs into segments short enough to sequence and call. This was through a choice of PCR conditions that they have reported previously [Kennedy et al. 2003], which exploit PCR's inherent size sensitivity to amplify fragments of the desired size and exclude others.

Because any superiority of SNPs over microsatellites is supposed to rest on the greater density of SNPs across the chromosomal landscape, Kennedy wanted to check that this indeed explained her own team's particular win. Returning for a reanalysis, she removed all but 3,300 roughly evenly spaced SNPs from the team's map terrain—a handicap that put their set of single nucleotides on a par with the 350 more variable microsatellites. This stripped-down SNP set indeed failed to spotlight the four new loci that the statistics of the full set had identified. Furthermore, it did implicate all of the loci that had emerged from the microsatellite analysis, while causing no entirely new ones to rise to a level of statistical significance—suggesting SNPs do not carry any more risk of false positives than do microsatellites, byte for byte. From their battery of tests Kennedy and John conclude that SNPs perform as advertised.

Although they hasten to add that the new loci remain hypotheses until pursued biologically, these are hypotheses, Kennedy points out, that would not had been formed without SNPs. A former microsatellite devotee, Kennedy now feels prepared to say of the older analytical option: "I wouldn't bother."

"If you don't have pedigrees and you have unrelated individuals," Kennedy says, "why wouldn't you use the markers that extract more information?"