Performances of single-SNP and pathway-based analyses of genome-wide data to detect genetic factors shared by eosinophil and basophil counts in asthma-ascertained families. C. Loucoubar1,2,3, M. Brossard1,2,3, PE. Sugier1,2,3, A. Vaysse1,2,3, P. Jeannin1,2,3, MH. Dizier1,2,3, M. Lathrop3, E. Bouzigon1,2,3, F. Demenais1,2,3, EGEA collaborative group 1) INSERM U946, Paris, France; 2) Université Paris Diderot, Paris, France; 3) Fondation Jean Dausset-CEPH, Paris, France.
Eosinophils (EOS) and basophils (BASO) are white blood cells (WBC) that coordinately mediate allergic inflammation. Correlation of their blood counts suggests the existence of shared genetic determinants. Genome-wide association studies of WBC counts in population-based studies have identified a number of loci and loci shared by EOS and BASO were reported in one study. Pathway-based methods have been proposed to facilitate the detection of trait-associated genes and these methods may be useful to identify pleiotropic genes. Our goal was to compare the ability of single-SNP and pathway-based analyses to detect genetic factors shared by EOS and BASO counts in 388 asthma-ascertained families from the French Epidemiological study on the Genetics and Environment of Asthma (EGEA) study. These families included 1,660 subjects who had been genotyped using an Illumina 610K chip. We used linear regression adjusted for family dependence to conduct single-marker association analysis of EOS and BASO counts with 2.5 millions Hapmap2-imputed SNPs. The pathway analysis,implemented in ALIGATOR, searched for overrepresentation of gene ontology(GO)classes containing association signals. The single-SNP analysis showed that the top hits were specific to EOS or BASO counts; the most significant loci, which represented new signals, were on 1p21.3 (p=2x10-7)for EOS and on 8q21.11 (p=4x10-6)for BASO. However, there were 7 loci(1q41, 2q36.1, 4q28.3, 6p21.3, 13q21.1, 15q14, 18p11.32) that included SNPs showing association signals (2x10-4p8x10-4) with both EOS and BASO and having the same direction of effects. The pathway based-analysis examined association results from 711,681 SNPs that lied within 18,482 genes which were assigned to 7,070 GO categories. We found an overrepresentation of 7 and 10 GO categories for EOS and BASO respectively that achieved a category-specific p-value6x10-4. These GO categories differed between EOS (immune response and lung development pathways) and BASO (calcium and sodium-potassium channel activities) and did not share any gene. This study shows that, in a situation where most trait-associated SNPs belong to genes that are spread across various biological pathways or are intergenic, pathway analysis does not have higher performance than single-SNP analysis to detect shared genetic determinants. Alternative gene-set methods to characterize pleiotropy will be further explored. Funding: ANR-11-BSV1-027-01 grant.
You may contact the first author (during and after the meeting) at