Genetic analysis and software development for studies on complex human traits

Abstract

Rapidly advancing genotyping and genome sequencing technologies have recently given us the opportunity to watch evolution at work. It turns out that the genome, like so many other workplaces, is rather messy. The sequences of no two humans are exactly alike, not even those of “identical” twins. A lot of this variability has allowed us as a species to adapt to different living conditions. But some of it leads to disease. This thesis consists of four studies aimed at elucidating the genetic mechanisms involved in heritable complex genetic disorders of the brain.

In paper I, we studied the involvement of genome-wide sequence variants of estrogen receptor (ER) binding sites in mood disorders. Based on previously reported gender differences we studied each gender separately and found that a polymorphism in the genetic code close to the gene TGM2 might be involved in the disease mechanism for bipolar disorder (BD) affecting women. During this study we developed several independent Perl programs, which could be useful also to other genetic researchers. Therefore, in paper III, we merged these programs together into one user-friendly tool called ReMo-SNPs and extended its features so it would be able to perform searches in any region or motif of interest genome-wide. By combining in silico identified binding motifs with experimentally validated in vivo or in vitro binding regions, the program enables the researcher to maximize the functionality and relevance of the identified SNPs.

In paper II we were interested in evaluating the possible involvement of copy number variations (CNVs) in BD, and schizophrenia (SZ). Based on linkage studies, we selected 227 individuals with high disease prevalence for genome-wide genotyping. CNV variation was assessed and a ~200 kilobase (kb) deletion in the MAGI1 gene was identified. Further analyses of CNVs in the MAGI1 and MAGI2 genes were performed in a pooled analysis comprising 10,925 BD or SZ cases and 16,747 controls. We identified eleven additional CNVs of >100 kb in the MAGI1 and the MAGI2 genes in cases, while only four CNVs were found in these genomic regions in the control group. An association analysis on the pooled sample resulted in a significant association (p-value 0.023). This supports a possible role for rare MAGI1 and MAGI2 CNVs as risk factors for BD and SZ.

In paper IV we tested associations of eight variations previously identified in GWAS in large European cohorts of migraine patients in a Swedish material. We found support for two of the previously identified SNPs, rs1835740 and rs2651899, which means that they might be involved in the disease mechanism also in Sweden.

Conclusions: The work performed in this thesis has increased our knowledge of the genetic mechanisms behind BD, SZ, major depression, and migraine, four complex genetic disorders. A SNP in an ER binding region close to the gene TGM2 might be involved in the disease mechanism for women with BD (paper I). Large (>100 kb), rare CNVs in the genes MAGI1 and MAGI2 might be involved in the disease mechanisms in BD and SZ (paper II). Our ReMo-SNPs program has become a useful tool to identify functional SNPs for association studies in any region or motif of interest genome-wide (paper III). Two SNPs, rs183740 and rs2651899, might be involved in the disease mechanism in Swedish patients with migraine (paper IV). Even though our findings only constitute a small piece in the gigantic puzzle of complex genetic disorders, they may be of help to future researchers, clinicians, and patients and bring us a few steps forward in our quest to sort out the messy workplace that is our genome.