Programmable Landscapes

Many structurally complex biomolecules are generated algorithmically, and not in a template-driven form. That is, they are synthesized by enzymes which implement a series of addition, subtraction and modification reactions in precise time order. The questions we ask are: what are the general limits of algorithmic assembly? What are the spectrum of errors which can arise from such a process? How precisely are the rules to assemble a given molecule stored in a genome? And finally, can we learn from biomolecular assembly to program the assembly of non-biological molecules for technological applications? We use ideas from self-assembly, rule-based systems and reactive systems to address these questions.

Many structures in biology are usefully represented as graphs. This includes protein interaction networks, dynamical systems corresponding to signalling systems, and trafficking networks by which cargo moves around within cells. Biologically-relevant graphs have special structural properties which are absent from "randomly generated" graphs. Starting with detailed data about molecules and their interactions, as determined by biochemists and molecular biologists, we try to abstrat general princples about the resulting graphs. In this way, it is possible to apply core graph-theoretic tests to check whether a given hypothetical graph is possible or impossible to generate from underlying biomolecular interactions. Remarkably, this approach allows us to rule out large numbers of hypotheses with very little knowledge of detailed biochemical interactions, but purely on the basis of general principles such as mass-balance, reversibility, and so on.

The membrane of a cell is an active, complex interface between the inside and outside of a cell. It is the cell's primary source of information about the external environment, and its spatial and temporal fluctuations. How best can we deploy molecular sensors across the membrane in order to sample these fluctuations? How can the resulting information be transmitted with least loss to the correct intracellular target? We use basic principles of Information Theory to formulate these questions in an abstract setting. This approach produces falsifiable predictions about the structure of the membrane and of the downstream signalling pathways.

Cell Fate Specification can be viewed as flow down an epigenetic landscape. How do these landscapes emerge from underlying gene regulatory circuits? We are using dynamical systems theory to parameterize and model these landscapes.