Bioinformatics Advance Access published May 22, 2007

mixing at both length scales (PCN and LIN) in these naturallyoccurring, evolutionarily selected, biological networks. Assortative mixing in LINs indicates that this feature in PCNs is independent of short-range interactions. The coefficient of assortativity (Newman, 2002), a measure of assortative mixing, are also found to be considerably high for both PCNs and LINs. By constructing appropriate control networks, we further demonstrate that the degree (connectivity) distribution of the PCNs alone can partially account for the presence of assortativity in these networks. To enumerate the contribution of these global parameters obtained from the coarse-grained network model of protein structures to their biophysical properties, we show that the coefficient of assortativity of PCNs and LINs tend to have positive correlation with the experimentally determined rate of folding of these proteins. This implies that assortative mixing, that tends to connect highlyconnected residues to other residues with many contacts, may assist in speeding up of the folding process. In contrast, the average clustering coefficients of LINs show a good negative correlation with the rate of folding, indicating that clustering of amino acids, that participate in long-range interactions, into cliques, slows down the folding process. Interestingly, the average clustering coefficients of PCNs show negligible correlation, thereby implying that the short range interactions can reduce the negative effect on their folding kinetics. Three parameters — CO (Contact Order) (Plaxco et al., 1998), LRO (Long Range Order) (Gromiha and Selvaraj, 2001), an