In this regard several existing computational pathway analys

In this regard, several existing computational pathway analysis methods can potentially identify deregulated signaling pathways from high-throughput datasets (Hung, 2013; Khatri et al., 2012). In particular, due to huge progress in microarray and sequencing techniques, computational methods tend to rely more on transcriptomic datasets to identify signaling pathways that differ between two conditions. However, these methods are generally not specific for identification of differentially active and sustained signaling that maintains the different phenotypes, but are more generic in nature to capture all possible deregulated signaling events that could also be transient in nature. On the contrary, our method is more specific, since it aims to identify only constantly activated/inhibited signaling pathways that are responsible for maintaining the phenotype-specific TRN state, and whose perturbations can destabilize this state. Indeed, sustained activation/inhibition of signaling pathways are shown to exhibit a clear influence on the expression of genes participating in such pathways, which is not always observed in cases of transient activation/inhibition (Whitmarsh, 2007). For example, gene expression signatures have been successfully employed to identify constantly active oncogenic addiction pathways (Bild et al., 2006). Furthermore, signaling pathways involved in differentiation and cellular growth are known to induce changes in expression of genes involved in the signal transduction (Codega et al., 2014; Zhu et al., 2006).
Results
Discussion Characterizing the regulatory relationship between stem cathepsin inhibitors and their niches is fundamental for understanding tissue homeostasis and its implications in disease conditions. Rapid strides are made in this direction by explicit experimental characterization of niche elements and their respective role in maintaining specific stem cell phenotypes (Scadden, 2014). A few computational methods relying on this information have been proposed to describe stem cell-niche interactions. However, the immense complexity that arises due to the many-body nature of stem cell-niche interactions, as well as niche dynamics, urges the implementation of alternative computational approaches. In this direction, we have developed a computational method that attempts to overcome this complexity by capturing the net effect of the niche that is reflected at the level of constant activation/inhibition of stem cell signaling pathways responsible for the stable maintenance of stem cell phenotypes. Further, we showed that our method, though simple in its framework, is able to capture experimentally confirmed niche determinants for different stem cell systems from a large set of DERs. Notably, from more than 300 DERs, our method was able to capture S1pr1 and Egfr for maintenance of neural stem cell quiescent and active states, respectively (Codega et al., 2014). Prediction of Cd44 as a top ranking receptor for maintaining active OPCs state is experimentally known in the context of OPC migration upon injury-induced activation (Piao et al., 2013). Similarly, in the case of activated SCs in response to injury, our method predicted Cd44, whose role in SC migration was observed experimentally (Kobielak et al., 2007). Interestingly, Cd44 mediated signaling seems responsible for injury-induced activation of both OPCs and SCs, and it might represent a common mechanism of stem cell response to injury. Further, it points to the potential application of our method for understanding degenerative disease mechanisms in general and can possibly aid development of novel therapeutic strategies. A precise quantitative assessment of overall sensitivity and specificity of our computational approach is not readily possible due to incomplete and ever increasing knowledge of niche mediated signaling pathways that regulate stem cell phenotypes. Particularly, the information about false positives and true negatives in stem cell systems are rarely available. However, in the case of false negatives, for some stem cell systems that were studied, we found that our approach could not identify certain known signaling pathways mediated by the niche. For instance, in the case of NSCs, prostaglandin signaling (mediated by Ptgdr) and Notch signaling are known to maintain the quiescent phenotype, and were not identified by our approach (Codega et al., 2014; Llorens-Bobadilla et al., 2015). Similarly, Wnt signaling which is known to play a role in long term maintenance of HSCs was not identified by our approach (Chotinantakul and Leeanansaksiri, 2012). Possible reasons for not identifying some of the known niche mediated signaling responsible for stem cell phenotype maintenance could be due to lack of information in the interactome databases, inherent noise in gene expression data, lack of good correlation between transcriptome and proteome/phosphoproteome levels and redundancy in signaling pathways. Despite these, our method, relying only on gene expression data could successfully identify several experimentally known candidates (true positives) of niche mediated signaling for maintaining specific stem cell phenotype.