Reevaluating SNP associations in view of genome structure

[编者的话]

关于SNP的研究，相信大家都很熟悉，我们一般用它来找寻疾病相关的位点。但是，也有研究者对这个想法提出疑问，在一个国际基因组药物会议上，他认为，仅仅利用简单的SNP的信息将得出错误的结论。

During the past year, scientists have found that recombination does not occur evenly across the human genome, as previously thought. Instead, recombination almost always occurs in small regions. In between these hot spots lie stable blocks of sequence that have undergone very little recombination during evolution. In fact, these recombination-poor blocks - or haplotypes, as they are called - span as much as 80% of human genomes.

This insight into the structure of the human genome is interesting in its own right, but it is astounding when considered in terms of disease-association studies, which are designed to identify polymorphisms, or SNPs, that segregate in a statistically significant manner with disease occurrence. The simplest interpretation in such cases is that the SNP used in the study is the one that causes the disease. But the implication of these new structural data is that within a stable block where all the SNPs will tend to segregate together, including those that have not yet been detected by researchers, the SNP used in the study effectively represents all the other polymorphisms in its block.

For example, in the case of the MDR1 gene, which plays a major role in drug availability and patient response to therapeutics, many researchers have found that drug resistance segregates with a particular SNP in exon 26 of that gene. From these observations, many scientists have concluded that this SNP itself is responsible for the drug-resistance phenotype.

But David Goldstein, chair of the department of genetics at the University of London, said today at the conference that his group has found MDR1 exon 26 to be in the middle of a large block of stable sequence. Thus the SNP commonly used in the association studies is simply a marker for any number of other SNPs that have yet to be identified in sequencing projects. "There is no reason to think that polymorphism is causal in the phenotype," Goldstein told BioMedNet News. "In fact, I would bet a large amount of money that it is silent" and not causal, though he declined to elaborate on his reasoning because the supporting data are not yet published.

With the newly revised view of genome structure, Goldstein and others at the conference suggest modifying the protocol for SNP association studies. According to their proposal, once researchers find a polymorphism that segregates with their disease of interest they should back up and look at the haplotype structure of the region - that is, define the stable blocks. Then, when they know which block contains their SNP of interest, they can sequence that region in a number of affected and nonaffected individuals to get a sense of what other polymorphisms exist and which one might be causing the phenotype.

Using a similarly modified SNP approach, Leena Peltonen, chair of the department of human genetics in the Geffen School of Medicine at the University of California, Los Angeles, told the conference that her group has identified SNPs in the non-coding intronic and regulatory regions of several genes - regions that ordinarily wouldn't have been examined.

In a gene that controls lactose intolerance, Peltonen found two SNPs associated with the phenotype, one located eight kb upstream of the coding region and another 14 kb upstream. Interestingly, she says, these SNPs are prevalent across very diverse human populations, indicating that they are ancient alleles and that lactose tolerance is conferred by a less common and new allele that is really a mutant variety. "I think this gives us a beautiful perspective on 'disease' genes," she says. "They have been beneficial in evolution."