Whole genome sequencing study of identical twins discordant for psychosis.

Monozygotic (MZ) twins are often thought to have identical genomes, but recent work has shown that early post-zygotic events can result in a spectrum of DNA variants that are different between MZ twins. Such variants may explain phenotypic discordance and contribute to disease etiology. Here we performed whole genome sequencing in 17 pairs of MZ twins discordant for a psychotic disorder (schizophrenia, schizoaffective disorder or bipolar disorder). We examined various classes of rare variants that are discordant within a twin pair. We identified four genes harboring rare, predicted deleterious missense variants that were private to an affected individual in the cohort. Variants in FOXN1 and FLOT2 would have been categorized as damaging from recent schizophrenia and bipolar exome sequencing studies. Additionally, we identified four rare genic copy number variants (CNVs) private to an affected sample, two of which overlapped genes that have shown evidence for association with schizophrenia or bipolar disorder. One such CNV was a 3q29 duplication previously implicated in autism and developmental delay. We have performed the largest MZ twin study for discordant psychotic phenotypes to date. These findings warrant further investigation using other analytical approaches.

Investigating copy number variants in schizophrenia pedigrees using a new consensus pipeline called PECAN.

Copy number variants (CNVs) have been implicated in many human diseases, including psychiatric disorders. Whole genome sequencing offers advantages in CNV calling compared to previous array-based methods. Here we present a robust and transparent CNV calling pipeline, PECAN (PEdigree Copy number vAriaNt calling), for short-read, whole genome sequencing data, comprised of a novel combination of four calling methods and structural variant genotyping. This method is scalable and can incorporate pedigree information to retain lower-confidence CNVs that would otherwise be discarded. We have robustly benchmarked PECAN using gold-standard CNV calls for two well-established evaluation samples, NA12878 and HG002, showing that PECAN performs with high precision and recall on both datasets, outperforming another pedigree-based CNV calling pipeline. As part of this work, we provide a list of high-confidence gold standard CNVs for the NA12878 reference sample, curated from multiple studies. We applied PECAN to a collection of pedigrees multiply affected with schizophrenia and identified a rare deletion that perfectly co-segregates with schizophrenia in one of the pedigrees. The CNV overlaps the gene PITRM1, which has been implicated in a complex phenotype including ataxia, developmental delay, and schizophrenia-like episodes in affected adults.

Ultrarare Missense Variants Implicated in Utah Pedigrees Multiply Affected With Schizophrenia.

Background: Recent work from the Schizophrenia Exome Sequencing Meta-analysis (SCHEMA) consortium showed significant enrichment of ultrarare variants in schizophrenia cases. Family-based studies offer a unique opportunity to evaluate rare variants because risk in multiplex pedigrees is more likely to be influenced by the same collection of variants than an unrelated cohort. Methods: Here, we examine whole genome sequencing data from 35 individuals across 6 pedigrees multiply affected by schizophrenia. We applied a rigorous filtering pipeline to search for classes of protein-coding variants that cosegregated with disease status, and we examined these for evidence of enrichment in the SCHEMA dataset. Additionally, we applied a family-based consensus approach to call copy number variants and screen against a list of schizophrenia-associated risk variants. Results: We identified deleterious missense variants in 3 genes (ATP2B2, SLC25A28, and GSK3A) that cosegregated with disease in 3 of the pedigrees. In the SCHEMA, the gene ATP2B2 shows highly suggestive evidence for deleterious missense variants in schizophrenia cases (p = .000072). ATP2B2 is involved in intracellular calcium homeostasis, expressed in multiple brain tissue types, and predicted to be intolerant to loss-of-function and missense variants. Conclusions: We have identified genes that are likely to increase schizophrenia risk in 3 of the 6 pedigrees examined, the strongest evidence being for a gene involved in calcium homeostasis. Further work is required to examine other classes of variants that may be contributing to disease burden.

Identity-by-descent analysis of a large Tourette's syndrome pedigree from Costa Rica implicates genes involved in neuronal development and signal transduction.

Tourette Syndrome (TS) is a heritable, early-onset neuropsychiatric disorder that typically begins in early childhood. Identifying rare genetic variants that make a significant contribution to risk in affected families may provide important insights into the molecular aetiology of this complex and heterogeneous syndrome. Here we present a whole-genome sequencing (WGS) analysis from the 11-generation pedigree (>500 individuals) of a densely affected Costa Rican family which shares ancestry from six founder pairs. By conducting an identity-by-descent (IBD) analysis using WGS data from 19 individuals from the extended pedigree we have identified putative risk haplotypes that were not seen in controls, and can be linked with four of the six founder pairs. Rare coding and non-coding variants present on the haplotypes and only seen in haplotype carriers show an enrichment in pathways such as regulation of locomotion and signal transduction, suggesting common mechanisms by which the haplotype-specific variants may be contributing to TS-risk in this pedigree. In particular we have identified a rare deleterious missense variation in RAPGEF1 on a chromosome 9 haplotype and two ultra-rare deleterious intronic variants in ERBB4 and IKZF2 on the same chromosome 2 haplotype. All three genes play a role in neurodevelopment. This study, using WGS data in a pedigree-based approach, shows the importance of investigating both coding and non-coding variants to identify genes that may contribute to disease risk. Together, the genes and variants identified on the IBD haplotypes represent biologically relevant targets for investigation in other pedigree and population-based TS data.

Converting single nucleotide variants between genome builds: from cautionary tale to solution.

Next-generation sequencing studies are dependent on a high-quality reference genome for single nucleotide variant (SNV) calling. Although the two most recent builds of the human genome are widely used, position information is typically not directly comparable between them. Re-alignment gives the most accurate position information, but this procedure is often computationally expensive, and therefore, tools such as liftOver and CrossMap are used to convert data from one build to another. However, the positions of converted SNVs do not always match SNVs derived from aligned data, and in some instances, SNVs are known to change chromosome when converted. This is a significant problem when compiling sequencing resources or comparing results across studies. Here, we describe a novel algorithm to identify positions that are unstable when converting between human genome reference builds. These positions are detected independent of the conversion tools and are determined by the chain files, which provide a mapping of contiguous positions from one build to another. We also provide the list of unstable positions for converting between the two most commonly used builds GRCh37 and GRCh38. Pre-excluding SNVs at these positions, prior to conversion, results in SNVs that are stable to conversion. This simple procedure gives the same final list of stable SNVs as applying the algorithm and subsequently removing variants at unstable positions. This work highlights the care that must be taken when converting SNVs between genome builds and provides a simple method for ensuring higher confidence converted data. Unstable positions and algorithm code, available at https://github.com/cathaloruaidh/genomeBuildConversion.