Eric Siggia
Center for Studies in Physics and Biology
Laboratory of Theoretical Condensed Matter Physics
Publications
Projects
Software
Lab Members
Applications of statistical physics and dynamical systems to cellular biophysics and bioinformatics.
Thermal fluctuations are an inherent feature of the large protein assemblies that are the functional subunits of the cell. Condensed matter physics deals with polymers, gels and membranes with approximate theories that parameterize atomic interactions yet retain the basic features induced by thermal effects. The elasticity of ds DNA over the range of forces relevant to the cell can be quantitatively understood interms of two elastic parameters, including the various shapes due to supercoiling. Semi empirical theories are an apt level to describe the physical chemistry of protein sorting in the cell or the kinetics of RNA folding.The genome is more than a 'parts list', and rather more akin to an assembly manual, which directs when and where genes are to be expressed in response to signals conveyed by proteins. However decoding the regulatory part of the genome is more challenging than finding the protein coding regions, since protein binding sites are of order 10 bases (vs a hundred or more bases for a typical exon encoding a protein), and there are as yet no rules governing their organization. There are many analogies between statistical physics and problems of pattern recognition that arise when one attempts to decode the regulatory part of the genome. Natural language processing deals with similar problems, but employs different means since the corpus of training data is incomparably greater.
There has been a very fruitful convergence in the biological literature between evolution and development leading to the paradigm that much evolutionary innovation, such as seen in development occurs through changes in gene regulation, rather than the creation of new genes. Thus it is informative to compare the regulation of related species, both to isolate the important regulatory elements based on their conservation, and to understand the structural constraints on regulatory modules. Formal theories as to how machines learn rules from examples may point to the types of logical operations regulatory modules have 'learned' to preform based on the examples supplied during evolution.