Our computational biology group believes in a new wave of scientific enlightenment in biology fuelled by intelligent software and predictive analytics.
Intelligent software is adaptive, scalable, and user-friendly. Inspired by recent advancements in deep learning and cloud-computing, we develop intelligent open-source software and harness it to translate the predictive capacity of artificial intelligence into molecular biology research. Our ultimate aim is to predict the regulatory evolution of gene expression and causally associate molecular mechanisms of gene regulation with phenotypic changes in complex traits.
We approach this by integrating comparative and functional genomics at tree-of-life scale with evolutionary transcriptomics and causal inference of gene regulatory networks to derive a data-driven predictive framework of trait evolvability.
We successfully applied aspects of this research strategy to a diverse portfolio of biological questions and keep refining this process to accommodate an increasing amount of incoming data and sequencing technologies and to serve a wider range of molecular biological applications. You can find a more detailed summary of our Scientific Software and previous research question on the Biological Research Portfolio pages.
The evolution of gene regulation
Natural variation is a crucial property of life because it is the outcome of a dynamic genome which creates vast combinatorial population diversity as a hedge against natural selection. Natural variation denotes individual variants of the population with the molecular composition to either adapt to changing environmental conditions or perish due to a lack of adaptive potential. It has been debated for decades whether point mutations, transposons, gene/genome duplications, or particular changes in gene regulation (e.g. pathways or gene regulatory networks) are the major molecular mechanisms generating natural variation and thereby causing adaptive processes in genomic landscapes (transgenerational genome dynamics).
To contribute to this debate about the molecular determinants of plasticity and adaptability, we use a comparative approach at tree-of-life scale to unveil universal principles of eukaryotic gene regulation and aim to causally associate changes in gene regulatory interactions at pathway scale with the adaptive evolution of complex traits.
As a result, this approach allows to ask question such as how are gene regulatory instructions retained across long evolutionary time scales and diverse species despite the presence of natural variation and heterochrony (shifts in developmental timing) driving overall changes in gene expression?
Detection of conserved and variable modules in gene regulatory networks using a comparative approach
To model this adaptive change on the organism-level, we use the reductionist concept of system-scale gene regulatory networks (sysGRNs) to quantify gene regulatory changes via changes in the topology of the inferred network graph. Comparing such sysGRNs across cell-types, individual organisms, ecotypes, and species enables the identification of conserved and variable sub-networks (= modules) that can be associated with molecular or phenotypic change and allows to identify entire biological pathways driving the evolution of a complex trait. In addition, we seek to trace the birth, sequence divergence, and copy-number-variation of each gene in the sysGRN to understand the evolutionary dynamics of [ regulator – target ] associations which may explain the emergence of adaptability via macro-evolutionary leaps caused by massive regulatory rewiring events.
For this challenging task, we build the mathematical, computational, and comparative frameworks to quantify module retention and adaptive change in sysGRNs and rely on in-house generated as well as publicly available transcriptome datasets to infer such networks covering various biological processes at various resolutions of organismal complexity.
Finally, our group seeks to train and mentor the next generation of interdisciplinary system-scale thinkers able to integrate insights from computer science, statistical modeling, and diverse life science disciplines to derive a macro-level understanding of natural variation and heredity. If you wish to join or otherwise support our efforts, please read further details here about how to get on-board.