Rapid development in technologies to read genomes has opened up unprecedented possibilities to discover links between our genomes and our health, quantitative traits and genetic ancestry. However, the massive amount of genomics data and limited understanding of its meaning at the level of individual make these goals challenging. To push the field forward we need truly interdisciplinary teamwork across medical, biological and computational sciences. Novel quantitative methods and new generations of quantitative scientists are a crucial part of continuing success in human genomics.
Our goal is to answer biologically and medically important questions quantitatively using data from large Finnish cohorts and international collaborations. This requires 1) understanding the properties and context of the data, 2) identifying appropriate statistical approaches and 3) designing computational implementations that work in practice. Typically our main challenge is to determine a balance between the level of complexity of the statistical models and their computational tractability. We publish both new statistical methods as well as their applications on large data sets.
Genome-wide association studies (GWAS)
A genome-wide association study (GWAS) maps statistical association between genetic variants and a quantitative trait (e.g. height or cholesterol levels) or a disease (e.g. migraine or multiple sclerosis). Our group has taken part in dozens of international GWAS, and we are regularly teaching the statistical methodology used in GWAS. Currently, our group is leading the analysis efforts of a global migraine GWAS that combines data from over 100,000 migraine cases and 750,000 population controls to learn about genetic susceptibility to migraine.
Multivariate statistical methods in human genetics
Over the last decade, numerous genetic regions have been associated with complex diseases and traits through GWAS. An exciting opportunity for quantitative research is to take these publicly available discoveries and tease out more from them by using multivariate approaches. We have recently made software, e.g., for multi-SNP (FINEMAP), multi-trait (metaCCA, MetaPhat) and multi-tissue data (ASE models).
In addition to large publicly available international datasets, we also have exciting projects in Finland to link detailed genetic data to hundreds of metabolites in tens of thousands of individuals.
Population genetics and distribution of genetic risk in Finland
Finland is known for its unique genetic background shaped by strong genetic drift from other European populations as well as between different parts of the country. With novel haplotype-based methods we can detect fine-scale genetic structure within Finland at an unprecedented scale. This way we lay a foundation for controlling confounding in rare variant association studies in Finland, for assessing genetic risk for diseases across Finland and for individual-level ancestry estimation within Finland. Our comprehensive collection of Finnish genetic data is a treasure for novel population genetics applications. As an example, we participated in making a special stamp for Finland's 100th anniversary.