Brunak Group (Translational Disease Systems Biology Group)
Group Leader: Prof. Søren Brunak, Research Director
The huge diversity of the human proteome is based on a genome that contains less than 20,000 basic protein coding genes. Many of these genes and proteins are likely involved in more than one disease, or in different subphenotypes of a group of diseases that link to a particular organ or tissue. It is, thus, difficult to translate genome and proteome variations to reliably predict the resulting phenotypes. We address these challenges by using patient-specific healthcare data that contain information on how disease combinations manifest themselves and co-occur in individual patients. Using the Danish personal identification system, we also simultaneously analyze disease co-occurrences in families to help understand how genes and protein complexes may be responsible for the same, or very similar, phenotypes.
The aim of the Brunak group is to identify new drug targets, biomarkers, and to provide diagnostic evidence, by:
- developing data integration tools in supercomputing settings
- expanding our already extensive collaborative networks to involve the Capital Region of Denmark and Region Zealand, and clinical partners
- recruiting MD-PhD students who can add clinical and medical knowledge to our team of bioinformaticians, systems biologists, and computer scientists.
A key achievement of 2015 was publishing several papers (e.g. Ellesøe et al., Congenit Heart Dis, 2015) on congenital heart disease (CHD) subphenotypes and their concordance and discordance in Danish families. Hierarchical cluster analysis of data obtained from patient records (1163 families with CHD in the National Danish Patient Registry, a total of 3080 individuals) showed the familial co-occurrence of six distinct clusters of diagnoses. The discordant co-occurrences largely matched the number of overlapping genes already identified in knockout mice as being associated with the co-occurring pair of subphenotypes. More than a 100 exomes obtained from blood samples were sequenced to understand the distribution of causal variants in protein complex components and many new potential CHD genes were identified.
We have developed a method to facilitate the interpretation of the consequences of human protein kinase variation. Using a kinase-specific random forest approach, we integrated nine methods that predicted the pathogenicity of variants. The variants were visualized in their structural contexts and residues affecting catalytic and drug binding were identified. Furthermore, we developed our disease trajectory concept by analyzing the temporal order and multiple paths of disease occurrence to estimate the mortality risk of individuals from a 20-year prehistory of a single diagnosis. This work resulted in a paper on sepsis (and the role of different diabetes types therein), and forms the basis for a new big data approach for aggregating long timescales combined with high frequency data from short timescales, e.g., from Intensive Care Units.
Researchers from the Brunak Group have published collaborative papers in Cell and Nature, including a paper disentangling type 2 diabetes (Forslund et al., Nature, 2015) and metformin treatment signatures in the human gut microbiota (Rasmussen et al., Cell, 2015).