ORLANDO, Fla. (GenomeWeb News) – At an American Association for Cancer Research annual meeting session here yesterday, attendees heard about some of the strategies that researchers are using to glean additional information from genome-wide association studies.
In particular, presenters highlighted the value of gene- and pathway-based analyses for helping to more rapidly gain biological and functional insights from genetic associations detected in these studies.
To that end, University of Chicago human geneticist Nancy Cox discussed ways in which expression quantitative trait loci — genetic variants associated with specific transcript levels — can be used to narrow in on SNPs and genes involved in disease, since studies suggest that eQTLs are enriched amongst the collection of SNPs linked to disease through GWAS so far.
Cox outlined some of the eQTL associations being identified in cancer, for instance, noting that these eQTLs seem to vary depending on the type of study being done, with pharmacogenomic studies apparently yielding more eQTLS related to distant transcripts and expression-related risk variants for solid cancers showing more local affects on expression.
She also discussed functional genome-wide association studies being used to study breast cancer and other diseases.
To do these functional GWAS, Cox explained, researchers sift through SNPs and only include those with known functional roles in their analyses. These may include eQTLs, missense, or nonsense mutations in protein-coding regions, and so on, she added.
Consequently, such an approach yields associations at the gene rather than SNP level, Cox said, providing immediate clues for follow-up functional studies. Still, she noted, the approach can be expanded to include functionally informative SNPs in non-gene regions as well — and for analyses of sequence data.
Meanwhile, Hakon Hakonarson, director of the Children's Hospital of Philadelphia Center for Applied Genomics, spoke about pathway-based analyses of large datasets from cancer and other disease GWAS.
Providing examples from inflammatory bowel disease and neuroblastoma studies, Hakonarson touched on the insights that can be gained from pathway analyses as well as the information that needs to be considered when doing these analyses.
After mapping SNPs to specific genes and pruning these SNPs down, he explained, accurate pathway analyses require researchers to calculate gene-based test statistics and pathway enrichment statistics and to adjust for both gene and pathway sizes.
Queensland Institute of Medical Research statistical genetics and genetic epidemiology researcher Stuart MacGregor offered additional insights into gene- and pathway-based analyses during the session, describing an interactive, simulation-based computational approach that he and his team developed to bring together information for SNPs across genes to get statistical information about gene associations.
The "versatile gene-based association study," or VEGAS method, has already been applied to several cancer datasets, MacGregor said, in some cases identifying genes that were not statistically significant in individual studies.
He also described a related tool known as VEGAS pathway that's already been used to explore the pathways underlying pancreatic, breast, and other cancers.
"Our approach … is applicable to all GWAS designs, including family-based GWAS, meta-analyses of GWAS on the basis of summary data, and DNA-pooling-based GWAS, where existing approaches based on permutation are not possible, as well as singleton data, where they are," MacGregor wrote in an abstract for the meeting.
Peter Holmans, a biostatistics and genetic epidemiology researcher at Cardiff University School of Medicine, meanwhile, touched on some statistical issues related to pathway-based studies and described his own pathway analysis approach.
The tool, called ALIGATOR, incorporates information about significant associations at the gene
level and connects genes from related pathways, Holmans said, helping to track down pathways that are enriched in specific diseases.
"If the susceptibility genes act together in a biological pathway, then a joint analysis of the set of genes in that pathway may increase power to detect association, and improve reproducibility across studies," Holmans wrote in the talk's abstract.