Identification of genetic risk variants for major psychiatric disorders in Costa Rican families using wgs

Abstract

One of the challenges in the study of complex disorders, including psychiatric disorders, is identifying genetic variants with varying frequencies and effect sizes. The well-known phenomenon of missing heritability in complex disorders can be explained in part by the difficulties in the identification of rare risk variants. Pedigree-based studies are a useful tool for the detection of rare genetic effects. In the present study, whole genome sequences were generated for 214 Costa Rican individuals belonging to families with high prevalence of different psychiatric phenotypes (one large multigenerational pedigree with 100 sequenced individuals, and 32 small families). Libraries were constructed and sequenced on the Illumina HiSeqX with the use of 151-bp paired-end reads for whole-genome sequencing. Samples were sequenced to a minimum of 30x average coverage. Variant calling was done using Isaac Genome Alignment and Isaac Variant Caller; ANNOVAR was used for variant annotation. Our genome-wide association analysis employed an additive measured genotype, fixed-effect mixed model variance component approach using genotype dosage as the focal covariate and allowing for non-independence amongst pedigree members. Covariates such as age, age2, sex, their interactions, and principal components of SNP variation to control for latent population structure were included in our model. The association analysis was performed for 4 different phenotypes: bipolar disorder, major depressive disorder, mood disorder, and psychosis. A genome-wide significant association with bipolar disorder was found for variants in the NRG1 and ERBB4 genes. For major depression, variants in several genes reached genome-wide significance (RNF115, SPATA45, FLVCR1, UBR3, NTSR2, IQCM, ENOX1, AIF1L, PTPRD, NTRK2, EXOSC2, AFAP1L2, GPC6, DYNLRB1, FMN2, GNG7, ITCH, MAPK1). For the mood disorder phenotype, there is one significant signal not located inside a gene. In the case of the psychosis phenotype there are signals in the CDH12 and NEIL1 genes that do not reach genome-wide significance. We have confirmed the previously reported involvement of several genes in psychiatric disorders, and also identified new genes for further study and confirmation.