Reading an interesting, well written, scientific paper can be a genuine pleasure. But, possibly because I am a bad person, I enjoy even more seeing a scientific paper taken down. If you do too then Aaron Clauset’s (quite old, 2008) post is for you. The post reanalyses some data from a paper by Yu et al., who study interactions between protein molecules in a simple organism: yeast cells. Cells, both ours and yeast’s, are mini-computers constantly computing stuff, e.g., when to divide, when to make more or less protein etc. This is partly done through proteins interacting with each other, and Yu et al. study these interactions, in particular they go throughs 100os of proteins and see which ones interact with which other ones. This is important useful work and Yu et al. produced a lot of useful data. It tells us important stuff about how cells work, and protein interactions going wrong can cause diseases.
Yu et al. go on to claim “As found previously for other macromolecular networks, the connectivity or “degree” distribution of all three datasets is best approximated by a power-law.” A power law is a function like y(x) ~ xs, here the function y(x) varies as the s-th power of the variable x. Prof Clauset is not so impressed by this claim, and having read his post, I can see why. The technical details are in the post, but basically their data don’t really fall on the line of a power law, any power law fit is pretty poor.
This matters. If it is a power law this might give us a big hint to how this network of interactions evolved. But as it is apparently not a power law we should take that into account when trying to understand how the proteins function inside yeast and ourself. As the always quotable Albert Einstein said “Everything should be made as simple as possible, but not simpler.” (Of course this is typically easier to say than to do, especially if you are not Albert Einstein) The pattern of protein interactions is apparently not as simple as a power law. A shame but that is, quite literally in this case, life.