Alina Frolova | Search bioRxiv

bioRxiv2016

Comparing alternative pipelines for cross-platform microarray gene expression data integration with RNA-seq data in breast cancer

BackgroundAccording to major public repositories statistics an overwhelming majority of the existing and newly uploaded data originates from microarray experiments. Unfortunately, the potential of this data to bring new insights is limited by the effects of individual study-specific biases due to small number of biological samples. Increasing sample size by direct microarray data integration increases the statistical power to obtain a more precise estimate of gene expression in a population of individuals resulting in lower false discovery rates. However, despite numerous recommendations for gene expression data integration, there is a lack of a systematic comparison of different processing approaches aimed to asses microarray platforms diversity and ambiguous probesets to genes correspondence, leading to low number of studies applying integration.\n\nResultsHere, we investigated five different approaches of the microarrays data processing in comparison with RNA-seq data on breast cancer samples. We aimed to evaluate different probesets annotations as well as different procedures of choosing between probesets mapped to the same gene. We show that pipelines rankings are mostly preserved across Affymetrix and Illumina platforms. BrainArray approach based on updated annotation and redesigned probesets definition and choosing probeset with the maximum average signal across the samples have best correlation with RNA-seq, while averaging probesets signals as well as scoring the quality of probes sequences mapping to the transcripts of the targeted gene have worse correlation. Finally, randomly selecting probeset among probesets mapped to the same gene significantly decreases the correlation with RNA-seq.\n\nConclusionWe show that methods, which rely on actual probesets signal intensities, are advantageous to methods considering biological characteristics of the probes sequences only and that cross-platform integration of datasets improves correlation with the RNA-seq data. We consider the results obtained in this paper contributive to the integrative analysis as a worthwhile alternative to the classical meta-analysis of the multiple gene expression datasets.

Bioinformatics↗

bioRxiv2015

Distributed Bayesian Networks Reconstruction on the Whole Genome Scale

BackgroundBayesian networks are directed acyclic graphical models widely used to represent the probabilistic relationships between random variables. They have been applied in various biological contexts, including gene regulatory networks and protein-protein interactions inference. Generally, learning Bayesian networks from experimental data is NP-hard, leading to widespread use of heuristic search methods giving suboptimal results. However, in cases when the acyclicity of the graph can be externally ensured, it is possible to find the optimal network in polynomial time. While our previously developed tool BNFinder implements polynomial time algorithm, reconstructing networks with the large amount of experimental data still leads to computations on single CPU growing exceedingly.\n\nResultsIn the present paper we propose parallelized algorithm designed for multi-core and distributed systems and its implementation in the improved version of BNFinder - tool for learning optimal Bayesian networks. The new algorithm has been tested on different simulated and experimental datasets showing that it has much better efficiency of parallelization than the previous version. BNFinder gives comparable results in terms of accuracy with respect to current state-of-the-art inference methods, giving significant advantage in cases when external information such as regulators list or prior edge probability can be introduced.\n\nConclusionsWe show that the new method can be used to reconstruct networks in the size range of thousands of genes making it practically applicable to whole genome datasets of prokaryotic systems and large components of eukaryotic genomes. Our benchmarking results on realistic datasets indicate that the tool should be useful to wide audience of researchers interested in discovering dependencies in their large-scale transcriptomic datasets.

Bioinformatics↗