What do we do?
Our main research interest at the Genome Systems Biology (GSB) Group is the study of genome-wide regulatory systems in order to reconstruct them from high-throughput molecular data, understand and model how they have evolved, and search for design principles in their construction. In particular, we are developing and applying new algorithmic tools for the automated reconstruction of genome-wide regulatory networks from comparative genomic, deep sequencing, and other high-throughput data. In addition, methods are being developed for studying genome evolution and the evolution of regulatory networks in particular.
Highlights 2016Our first main highlight for 2016 is the completion of our integrated microfluidic and image-analysis setup for the studying gene regulation in vivo at the single-cell level. Our setup, consisting of a dual-input microfluidic chip and accompanying analysis software allows automated and highly accurate tracking of the growth and gene expression of lineages of single cells as they respond to continuously changing external conditions. The analysis software, which we developed in collaboration with the group of Gene Myers, jointly optimizes both segmentation and tracking and includes a highly novel curation procedure, called `leveraged editing', in which wherein a single input directive can fix up to a dozen errors. Applying this methodology to the founding system of studies in gene regulation; induction of the lac operon in response to a switch of carbon source from glucose to lactose, we discover that single-cell lag times have a multi-modal distribution and that lag times are controlled by an (as of yet unknown) heritable factor.
A second highlight is the publication of our manuscript on our CRUNCH pipe-line for completely automated analysis of ChIP-seq data, which includes all steps from the quality analysis and mapping of the raw reads, up to comprehensive de novo motif finding and annotation of binding sites in all ChIP peaks. Finally, this year we also completed our development of a new general motif model, called Dinucleotide Weight Tensor (DWT), that incorporates arbitrary dependencies between positions within regulatory sites. In our recently submitted manuscript we show that DWTs, which have no tunable parameters whatsoever, always perform at least as well as position-specific weight matrices, and atrongly outperform them for a substantial fraction of transcription factors.
Main publications 2016
- Berger S et al. Crunch: Completely Automated Analysis of ChIP-seq Data. bioRxiv 042903; 2016
- Kaiser M et al. Tracking single-cell gene regulation in dynamically controlled environments using an integrated microfluidic and computational setup. bioRxiv 076224; 2016
- Omidi S, van Nimwegen E. Automated Incorporation of pairwise dependency in transcription factor binding site prediction using dinucleotide weight tensors. bioRxiv 078212; 2016.