Thus we confirmed the role of quantitative PTEN protein expression as a key determinant and putative biomarker of therapeutic resistance. One of the major barriers to more successful translation of I-BET-762 solubility dmso the results of modelling studies into clinical practice and anti-cancer drug development is a high level of individual variability of the cellular networks involved in seemingly identical cancers, not only due to genomic abnormalities (Kan et al., 2010), but also complex post-transcriptional and post-translational variability

in protein signalling networks (Faratian et al., 2009a). This causes a significant variation in individual responses to targeted anti-cancer treatments and therefore questions the practical utility of conclusions that can be drawn from network models with fixed parameters. Indeed, the majority of existing cancer-related modelling studies have been performed selleck products in a canonical way, where network model construction is followed by its parameterisation

via fitting the model to experimental data, and further analysis of one or several best solutions (Birtwistle et al., 2007, Chen et al., 2009, Faratian et al., 2009b and Schoeberl et al., 2009). The experimental data, used for model calibration, usually represent a set of time-course profiles of changes in protein phosphorylation, observed in response to perturbation of signalling with various receptor ligands. Given that such data are normally registered

for a particular cancer cell line, the quantitative predictions (e.g. on promising drug targets) drawn from the model analysis, though applicable to the reference cell type, may not be readily transferable TCL to other subtypes of cancer, due to possible biological variation of the network parameters in different cell lines, as well as potential noise in parameter estimates caused by the noise in experimental data. This may explain the slow incorporation of systems biology approaches as credible clinical tools. Another key but related impediment is the non-identifiability of model parameters, a problem common to many large-scale network models (Chen et al., 2009, Hengl et al., 2007, Rodriguez-Fernandez et al., 2006 and Yue et al., 2006). In complex biochemical models many parameters remain uncertain even when additional data are generated and different fitting algorithms are implemented (Brown and Sethna, 2003 and Chen et al., 2009). The majority of modelling studies employ various types of sensitivity analysis (SA) to assess how variation in input parameters can affect the model output. The most generally used method is local sensitivity analysis (LSA), based on evaluation of the impact of single parametric perturbations on the model output in close proximity to a reference solution, defined by nominal parameter values.