NEW YORK (GenomeWeb News) – A team from the Universities of Massachusetts and Pennsylvania is finding new contributors to genetically complex common diseases or traits using a statistical strategy that harvests locus-level information from existing genome-wide association study data.
The investigators presented their method — known as "Mixed modeling of Meta-Analysis P-values," or MixMAP — in PLoS ONE study this week. There, they used MixMAP to interrogate SNP information from GWAS datasets generated through past cholesterol-related projects, unearthing a dozen previously undetected gene-level loci now suspected of having ties to low-density lipoprotein cholesterol levels in the blood.
Unlike analyses based on individual SNP level information in GWAS data, members of the team explained, the MixMAP approach also considers genome structure and looks for associations occurring at the locus level.
In the iteration of the approach applied in the current study, for instance, MixMAP focuses on potential associations for gene level loci. Even so, authors explained that this method could be extended to more loci as they become better annotated. Moreover, they noted that the method may eventually be applied to try to find additional risk variants for a wide range of well-studied conditions.
"[O]ur method is straightforward to use with freely available computer software and can be applied broadly to advance genetic knowledge of many diseases," first author Andrea Foulkes, director of the University of Massachusetts at Amherst's Institute for Computational Biology, Biostatistics and Bioinformatics, said in a statement.
"We hope this moves us toward greater understanding of common disorders and improving overall health in our society," she continued.
The MixMAP strategy stemmed from an interest in more completely understanding the heritability of complex diseases. To that end, the team reasoned that there would likely be some benefit to looking at sets of variants in defined bits of the genome, rather than homing in on single SNPs showing significant ties to the trait or condition in question.
"While this [SNP-based] approach is valid," researchers explained, "we conjecture that substantial, complementary knowledge about association can be acquired by considering available information on all SNPs within a locus simultaneously in characterizing association."
The MixMAP method they designed to achieve that goal looks for such locus-level associations using p-values for SNPs assessed through past GWAS, which are newly analyzed using mixed effects modeling.
"The primary inputs required for this approach are single SNP level p-values for tests of trait association and mapping of SNPs to locus regions," they explained, "while the output is locus level estimates and tests of association."
For the current study, researchers gave their approach a go using publicly available LDL cholesterol SNP association data generated by the Global Lipids Gene Consortium as well as SNP data generated for the Penn Coronary Artery Calcification study using the ITMAT-Broad-CARe 50K SNP array.
Using its newly developed algorithm, for instance, the group saw associations at 21 of the 26 loci previously reported by the Global Lipids Gene Consortium.
But investigators also identified another 12 sites in the genome not highlighted by that LDL cholesterol association study. Among them were genes suspected of participating in pathways related to lipid or lipoprotein metabolism.
From the smaller Penn Coronary Artery Calcification dataset, meanwhile, the MixMAP analysis uncovered seven genes with apparent LDL cholesterol associations. These fell at loci lacking individual SNPs with significant ties to LDL cholesterol levels in that dataset, though researchers noted that individual variants at two of the loci did show ties to the lipid trait in the Global Lipids Gene Consortium data.
Though they warned that more research will be needed to shore up new gene associations — and delve into their potential biological significance — study authors say MixMAP appears to be a promising tool for gleaning new insights from datasets already analyzed via standard, SNP-focused methods.
"MixMAP offers novel and complementary information as compared to SNP-based analysis approaches," they wrote in PLoS ONE, "and is straightforward to implement with existing open-source statistical software tools."