There is an increasing recognition that complex systems are shifting their structure and behaviour between two major states: a more plastic and a more rigid state. Network plasticity increases the learning potential of the system. However, a highly plastic system will be unable to keep changes: will not have memory. An increase in system rigidity increases the memory storing ability of the system. Alternating changes of plasticity and rigidity emerge as a highly efficient optimization strategy. Importantly, rapidly dividing cells have more plastic networks than differentiated cells. As an example of this, the community structure of the protein-protein interaction network of yeast cells became more condensed upon stress (PLoS Comput. Biol. 7, e1002187). Plastic networks are dissipating perturbations well. (Actually, too well…) Therefore, plastic networks need to be attacked at their most rigid segments, i.e. at their most central nodes. Importantly, this is the drug targeting strategy of rapidly growing cancer cells or parasites. On the contrary, if rigid networks are attacked at their most central nodes, they easily become over-saturated, over-excited. This may lead to increased side-effects and toxicity. Therefore, rigid networks need to be attacked at their most plastic segments, which are in the neighbourhood of the most central network nodes. Importantly, this is the drug targeting strategy of differentiated cells in all other diseases (Pharmacology & Therapeutics, 138, 333-408).
Earlier several node-types, such as hubs, bridges, bottlenecks have been proposed as key actors of cellular processes. Our group developed the ModuLand program package, which is a novel method-family detecting pervasively overlapping communities, as well as their core-node(s), bridges and ‘creative nodes’ connecting multiple modules at the same time (www.modules.linkgroup.hu). Inter-community nodes play a critical role in the regulation of networks, such as our high-confidence signalling network, www.SignaLink.org. Recently we defined two novel centrality measures based on the excellence in perturbation propagation (perturbation-centrality;www.turbine.linkgroup.hu), or on the ability of a node or edge to establish or break cooperation in repeated social dilemma games (game-centrality; www.NetworGame.linkgroup.hu). Both dynamic centrality types defined nodes with higher biological relevance than conventional network centrality measures. We developed a toolkit, which is a powerful algorithm to find multiple nodes (node-sets) as best seeds of network perturbations causing a pre-defined final network state, or maintaining network cooperation. We also established a method finding the most independent groups of a network. These novel algorithms are potentially important novel tools in multi-target and allo-network drug design.
Peter Csermely is a professor at the Semmelweis University (Budapest, Hungary). His major fields of study (www.linkgroup.hu) are networks, adaptation and ageing. In 1995 Prof. Csermely launched a highly successful initiative, which provided research opportunities for more than 10,000 gifted high school students so far (www.nyex.de). In 2006 he established the Hungarian National Talent Support Council (www.tehetsegpont.hu) running a talent support network involving approx. 200,000 people. In 2012 he became the chair of the European Council of High Ability working for a Europe-wide network of talent support (www.talentcentrebudapest.eu). He wrote and edited 13 books (including the Weak Links at Springer) and published 270 research papers with a cumulative impact over 670 and total independent citations over 7,000. Dr. Csermely was the member of the Wise Persons’ Council of the president of Hungary, is a member of the Hungarian Academy of Sciences and Academia Europaea and an Ashoka Fellow, was a Fogarty, a Howard Hughes and a Rockefeller Scholar, and received several other national and international honors and awards including the 2004 Descartes Award of the European Union.