At #5 is a team led by Michael Wigler, of the Cold Spring Harbor Laboratory in New York, while at #9 is a very similar paper by the team led by Charles Lee, Brigham and Women’s Hospital, Boston, Massachusetts. Both have identified previously unsuspected variation in the large-scale structure of the human genome. There has already been considerable interest in variation in individual letters of the genetic code, called SNPs, for single nucleotide polymorphisms (see Science Watch, 17[3]: 8, May/June 2006) and in shared patterns of much larger blocks, known as haplotypes (Science Watch, 14[1]: 8 January/February 2003, and 15[6]: November/December 2004). Wigler’s and Lee’s groups found variation at a scale between these two, in chunks of genetic code up to about 2 million letters long.

These differences are called LCVs, for large-scale copy-number variations. LCVs are obviously much longer than SNPs and are fundamentally different from some other kinds of variation such as tandem repeats, which are like stutters in the sequence. Lee’s group scanned 55 people and found 255 places in the genome where individuals differed by fairly large chunks. Of these, 102 occurred in more than one person and 24 of them were present in more than 10% of the people studied. Wigler’s group looked at fewer people, only 20, and found fewer LCVs, only 76. On average, individuals differed by 12.4 LCVs in the Lee study and by 11 in the Wigler study. Only 11 of the LCVs were common to both, but the differences between the studies are possibly related to the different populations and the different methods used to identify the LCVs and are probably not important.

Why do these differences matter? About 5% of the human genome is known to be duplicated, with widely separated segments sharing 90% of their sequence over a stretch of 1,000 bases or more. These duplicated areas can mistakenly recognize one another during meiosis, when chromosomes exchange material, and so permit large rearrangements of the chromosomes. They are thought to be important in ordinary evolution and in genetic diseases. Both teams report a physical association between the location of LCVs and the location of these duplicated regions on the chromosomes. There is also an association with regions that are linked to genetic diseases and cancer. This suggests that the same mechanism—a mismatch during meiosis—is the basis of LCVs and other rearrangements of the genome, including some of those responsible for disease.

Intriguingly, Lee’s group reports that many of the LCVs they isolated (almost 13%) are "missing" from the consensus human genome sequence; they fall in the known gaps in the sequence. This could be because the duplications were eliminated by the sequencing software, as being too confusing. Or perhaps the software decided that two sequences that were actually from different parts of the genome were one and the same. Either way, the discovery of LCVs further confuses the issue of just how representative the consensus sequence is.

In both studies LCVs overlap with known genes. Thus some individuals have more copies of a particular gene than others, but it is far too soon to say that LCVs are always associated with disease, or are always benign. Both are certainly possible. Regulatory mechanisms might counter the effects of multiple copies, ironing out the differences between individuals. In other cases, however, copy number definitely affects gene expression. Wigler, for example, found variation in, among others, genes involved in leukemia and drug resistance in breast cancer. And there seem to be copy number variants that protect against HIV infection (E. Gonzalez, et al., Science, 307[5714]: 1434–40, 2005) and against malaria (V. Thathy, et al., Malaria Journal, 4: 54, 2005). It may be that LCVs—hitherto with no observed consequences—might point researchers to areas of the genome that are inherently unstable and at which new diseases may be discovered. That, and the search for associations between LCVs and disease, may explain some of the citation interest in the two papers.