Query-based biclustering of gene expression data using Probabilistic Relational Models

Abstract

BackgroundWith the availability of large scale expression compendia it is now possible to view own findings in the light of what is already available and retrieve genes with an expression profile similar to a set of genes of interest (i.e., a query or seed set) for a subset of conditions. To that end, a query-based strategy is needed that maximally exploits the coexpression behaviour of the seed genes to guide the biclustering, but that at the same time is robust against the presence of noisy genes in the seed set as seed genes are often assumed, but not guaranteed to be coexpressed in the queried compendium. Therefore, we developed ProBic, a query-based biclustering strategy based on Probabilistic Relational Models (PRMs) that exploits the use of prior distributions to extract the information contained within the seed set.ResultsWe applied ProBic on a large scale Escherichia coli compendium to extend partially described regulons with potentially novel members. We compared ProBic's performance with previously published query-based biclustering algorithms, namely ISA and QDB, from the perspective of bicluster expression quality, robustness of the outcome against noisy seed sets and biological relevance.This comparison learns that ProBic is able to retrieve biologically relevant, high quality biclusters that retain their seed genes and that it is particularly strong in handling noisy seeds.ConclusionsProBic is a query-based biclustering algorithm developed in a flexible framework, designed to detect biologically relevant, high quality biclusters that retain relevant seed genes even in the presence of noise or when dealing with low quality seed sets.

Network inference via adaptive optimal design

“…Before getting into details, it is important to remark that network inference may aim at different levels of accuracy [6,7]. The lowest level approach leads to a relatively rough impression of the network structure.…”

“…If, therefore, a zero is present in more than one linearization point, the chances are increasingly high that there is indeed no connection between node i and j in the network structure. Our approach contributes to the field of systems biology where the recovery of networks based on stimulus-response data is a central topic that has already caught a lot of attention [6,7]. In systems engineering, model based experimentation has matured substantially but here the design is usually developed for recovery of a set of model parameters in the original non-linear model structure and not for network inference, see e.g.…”

Mining Local Staircase Patterns in Noisy Data

“…Many solutions have been proposed, to name just a few, (non-parametric) probabilistic inference [17], [19], statistical relational models [20]. However, these studies did not pay attention to the overlapping structure of the detected biclusters as our work.…”

Noise-Resistant Bicluster Recognition

“…S4VD [16] improved SSVD [15] by incorporating a stability selection technique. Recently, query-based biclustering algorithms [32], [33] are developed, which utilize a set of seed genes provided by the user to prune the search space and guide the biclustering algorithm. Hanczar et al [34] focus on the corrected measurement of the biclustering methods.…”