engleski

Novel map descriptors for characterization of toxic effects in proteomics maps

We consider a novel numerical characterization of proteomics maps based on construction of a graph obtained by connecting all protein spots in a proteomics map that are at distance equal to, or smaller than, a critical distance D_c. We refer to so constructed graph as cluster graph and we calculate four associated characteristic matrices, previously considered in the literature: (1) The Euclidean distance matrix ED¸ (2) The adjacency distance matrix AD ; (3) The path distance matrix based on the shortest paths between connected spots PD ; and (4) The quotient matrix Q, the elements of which are given as the quotient of the corresponding elements of ED and AD matrices. Numerical descriptors for proteomics maps include in particular the leading eigenvalue of the Q matrix and family of associated "higher order" matrices defined as powers of Q. These map descriptors show considerable sensitivity to perturbation of proteomics maps by toxicants.

graph-theoretical descriptors; matrix invariants; proteomics map; quotient matrix; proxime proliferators

nije evidentirano