Evaluation of Transcription Factor Activity in Gene Regulatory Networks

Introduction Comprehending gene regulation has been a major endeavour in the field of bioinformatics for many years and still poses a difficult problem with a lot of uncertainties. In spite of improvements in the technologies used, e.g. high-throughput methods like ChIP-sequencing, and progress in the analysis of the driving regulators, large parts of many gene regulatory networks still remain to be unveiled (Röttger et al., 2012). This is especially true for the human genome which is most interesting for us as detailed knowledge on regulatory networks helps us to further elucidate physiological processes. Naturally a diverse set of methods for broadening our understanding of gene regulation has been published over the years. I therefore propose to investigate and discuss a recently presented approach for estimating the activity of transcription factors (TFs) (Schacht et al., 2014) as it promises further progress in this area. Goals My first aim is to implement the method described by Schacht et al. (2014) and reconstruct each step necessary for this analysis of regulatory interactions as closely as possible to the original paper's specifications with the only exception being the underlying gene regulatory network. The network is an integral part of the published method, since the method aims to quantify the effects TFs have on the expression of networked genes. For this I will use a regulatory network suggested by Thomas et al. (2014). The data forming this network should be highly reliable as it was created by combining text mining with expert curation. Furthermore it is publicly available, in contrast to the data from the MetaCore database used by Schacht et al. After applying this method to the given network, I am interested in how well it succeeds in explaining gene regulation. Therefore I will compare it with a tool called ISMARA which also aims on elucidating regulatory interactions by identifying the most influential regulators. Analysing how many regulators both methods agree on may as a first step indicate its soundness. Additionally I will apply cross validation and compare results to literature findings to further evaluate the method in its capabilities to explain regulation. Approach Method Schacht et al. describe a method to measure TF activity by combining microarray data and an underlying regulatory network. Depending on the regulatory relationships in the network, in this approach the activity of a TF is calculated by measuring the expression of its affected genes. This has the advantage …