Sparsity in Everything: Noisy Protein-Protein Interaction

Last time, we talked about useful (aka non Junk) DNA as our first example in the "Sparsity In Everything" series. Today we will stay in the biology territory with another suggestion: nonfunctional protein-protein interactions (PPIs). According to Emmanuel Levy and Christian Landry, the pairwise interaction of different proteins leads to a high phase space but they suggest that a certain amount of these interactions are in fact similar to junk DNA i.e. non functional If you have a device looking at protein-protein interaction then the number of non-functional interactions would be a sparse set in the larger set of functioning PPIs. In that case, that device could use Compressive Sensing techniques to find that set in fewer measurements than a tedious one-to-one pairwise study. PPI screeniing can be performed using High Throughput Screening methods (HTS is being investigated by the folks at Michigan). Emmanuel Levy and Christian Landry's article is entitled: How Perfect Can Protein Interactomes Be?. The abstract reads:

Any engineered device should certainly not contain nonfunctional components, for this would be a waste of energy and money. In contrast, evolutionary theory tells us that biological systems need not be optimized and may very well accumulate nonfunctional elements. Mutational and demographic processes contribute to the cluttering of eukaryotic genomes and transcriptional networks with "junk" DNA and spurious DNA binding sites. Here, we question whether such a notion should be applied to protein interactomes—that is, whether these protein interactomes are expected to contain a fraction of nonselected, nonfunctional protein-protein interactions (PPIs), which we term "noisy." We propose a simple relationship between the fraction of noisy interactions expected in a given organism and three parameters: (i) the number of mutations needed to create and destroy interactions, (ii) the size of the proteome, and (iii) the fitness cost of noisy interactions. All three parameters suggest that noisy PPIs are expected to exist. Their existence could help to explain why PPIs determined from large-scale studies often lack functional relationships between interacting proteins, why PPIs are poorly conserved across organisms, and why the PPI space appears to be immensely large. Finally, we propose experimental strategies to estimate the fraction of evolutionary noise in PPI networks.

This article was found via the genomicron blog.

Credit Photo: Ralf Vandebergh: Neighbourhood of Maastricht Netherlands. Photograph taken from the ground of the International Space Station with the new truss and with the Space Shuttle attached two days ago.