Hotspots of large rare deletions in the human genome

We have examined the genomic distribution of large rare autosomal deletions in a sample of 440 parent-parent-child trios from the Quebec founder population (QFP) which was recruited for a study of Attention Deficit Hyperactivity Disorder.

DNA isolated from blood was genotyped on Illumina Hap300 arrays. PennCNV combined with visual evaluation of images generated by the Beadstudio program was used to determine deletion boundary definition of sufficient precision to discern independent events, with near-perfect concordance between parent and child in about 98% of the 399 events detected in the offspring; the remaining 7 deletions were considered de novo. We defined several genomic regions of very high deletion frequency ('hotspots'), usually of 0.4-0.6 Mb in length where independent rare deletions were found at frequencies of up to 100 fold higher than the average for the genome as a whole. Five of the 7 de novo deletions were in these hotspots. The same hotspots were also observed in three other studies on members of the QFP, those with schizophrenia, with endometriosis and those from a longevity cohort.

Nine of the 13 hotspots carry one gene (7 of which are very long), while the rest contain no known genes. All nine genes have been implicated in disease. The patterns of exon deletions support the proposed roles for some of these genes in human disease, such as NRXN1 and PARKIN, and suggest limited roles or no role at all, for others, including MACROD2 and CTNNA3. Our results also offer an alternative interpretation for the observations of deletions in tumors which have been proposed as reflecting tumor-suppressive activity of genes in these hotspots.