Figure 1

The difference between complex purifications and protein complexes. (a,b) When two proteins belonging to the same complex are tagged in independent yeast strains and the other components of the complex identified, the two purifications rarely return precisely the same list of components. (c,d) Although proteins in a complex are associated, it is usually unclear which proteins interact directly with each other. To predict direct interactions, either the matrix (c) or spoke model (d) is applied to lists of co-purified proteins. To evaluate such interactions Gavin et al. [1] invented the socio-affinity index (SAI). In brief, the SAI quantifies the tendency for a protein pair (for example, Aro1 and Gfa1) to identify each other when one of them is tagged (as in b) and to co-purify when other proteins are tagged (as in a) relative to what would be expected from their frequency in the dataset: that is, how many times this protein was found as prey. High-affinity SAI values result when both proteins co-purify when either one is tagged (without co-purifying many other proteins) and when both are always seen together in purifications made with other baits. (e) For this particular complex one core, two modules and two attachments have been identified. Note that modules cannot be computed from only two purifications; the assemblies of Figure 1e are derived from the eight purifications shown in Table 2 and additional purifications not shown.