From gene expression profile to network analysis

Pegah Khosravi

School of Biological Sciences, IPM, Tehran, Iran

Abstract

Cancer is a serious genetic disease and identifying regulators in cancer using novel systems biology approaches will potentially lead to new insight into this disease. We sought to address this by inferring co-expression networks and gene regulatory networks (GRN). Moreover, comparative topological analysis of co-expression networks and GRNs can explain how regulators change between different conditions, such as cancer subtypes.

In this study, we reconstructed independent co-expression networks and GRNs from each cancerous stage using current state-of-art reverse engineering approaches. Next, we highlighted pathways, GOs, crucial genes and interactions involved in prostate cancer by analyzing each network individually and also in comparison with other networks.

We presented a graph based approach to detect critical pathways and GOs related to prostate cancer that due to determining new pathways and GOs. Also, essential transcription factors were searched based on topological analysis of gene regulatory networks.

Moreover, we describes a novel algorithm, “Signing of Regulatory Networks” (SIREN), which can infer the regulatory type of interactions in a known gene regulatory network given corresponding genome-wide gene expression data.

Our results showed that SIREN is an efficient algorithm with low computational complexity; hence, it is applicable to large biological networks. We also indicated that comparative analysis via integrating wide range of relevant studies of biological networks may provide useful information to gain better understanding of the cell.

A quantum leap un the reproducibility precision, and sensitivity of gene-expression-profile analysis even when sample size is extremely small

Limsoon Wong
National University of Singapore

Abstract

Over the past decade, many methods have been proposed to find relevant disease-causing mechanisms from gene-expression data.To date, no method is able to reliably identify disease mechanisms in extremely-small-sample-size situations. Even in a moderately-large-sample-size situation, microarray analysis shows low consistency when applied to independent datasets of the same disease phenotypes.In this talk, we first dissect the reasons for the failure of some well-known pathway-based gene-expression analysis methods.Surprisingly, these methods fail for various rather fundamental reasons, including inappropriate null hypothesis, instability when sample size is small, and signal dilution from normal-behaving branches of large relevant pathways.Then, by logically fixing each issue directly, we show a quantum leap in the reproducibility, precision, and sensitivity of gene selection from gene-expression profiles, even when sample size is extremely small and even when there are significant batch effects in the sample.(This talk is based on the work of my student Kevin Lim.)

Interactome-transcriptome analysis provides clues to the discrepancies in male infertility genetic association studies

Naser Ansari-Pour

Faculty of New Sciences and Technology, University of Tehran, Tehran, Iran

Abstract

Identifying the underlying genetics of Y-linked male infertility including pinpointing causal variants in genes within the male specific region of Y (MSY) has been a quite challenging process with setbacks during the past decade. Since complex diseases result from the interaction of multiple genes and also display considerable missing heritability, network analysis is more likely to explicate their underlying molecular basis. A network medicine approach was undertaken to identify the role of Y in spermatogenic failure (SpF). Results showed that a number of Y genes were differentially expressed even though SpF patients were negative for Y microdeletions, suggesting that absence of microdeletions does not rule out involvement of Y. The reconstructed co-expression network of Y revealed RPS4Y2 as the hub gene. Interestingly, this gene was not present in the Y network reported in the interactome implying that interactome databases are yet to be fully completed and/or that this co-expression network is unique to the testicular tissue while that reported in the human interactome is an amalgamation of all human cells and tissues. I will address how the well-known genotype-phenotype correlation discrepancy found in Y-linked male infertility association studies (i.e. the red-herring case of USP9Y) can be explained using the re-constructed network. Finally, I will present how a Y-centric SpF network can explain more than half of the genetic heterogeneity observed in this complex disease.

Some Algorithms for Inferring Gene Regulatory Networks Using Gene Expression data

Changiz Eslahchi & Rosa Aghdam

Shahid Beheshti University & School of Biological Sciences, IPM, Tehran, Iran

Abstract

Inferring Gene Regulatory Networks (GRNs) from gene expression data is a major challenge in systems biology. Path Consistency (PC) algorithm is one of the popular methods in this field. However, as an order dependent algorithm, PC algorithm is not robust, because it achieves different network topologies if gene orders are permuted. In addition, the performance of this algorithm depends on the threshold value used for independence tests. Consequently, selecting suitable sequential ordering of nodes and an appropriate threshold value for the inputs of PC algorithm are challenges to infer a good GRN. In this work, we propose a heuristic algorithm, namely SORDER, to find a suitable sequential ordering of nodes. Based on SORDER algorithm and a suitable interval threshold for Conditional Mutual Information (CMI) tests, a network inference method, namely Consensus Network (CN), is developed. In the proposed method, for each edge of the complete graph, a weighted value is defined. This value is considered as the reliability value of dependency between two nodes. The final inferred network, by CN algorithm, contains edges with reliability value of dependency more than a defined threshold. The effectiveness of this method is benchmarked through several networks from DREAM challenge and the widely used SOS DNA repair network in Escherichia coli. The results indicate that the CN algorithm is suitable for learning GRNs and it considerably improves the precision of network inference. The source of data sets and codes are available at http://bs.ipm.ir/softwares/CN

Bridging the gap between expression profiling and functional inference: the enrichment analysis

(Introducing Enrichr, BinGO, DAVID and PEIMAN)

Mohieddin Jafari

Institute Pasteur, Tehran, Iran

Abstract

Despite high-throughput data helps to speed up biological data generation, inference of biological functions gets harder. Network biology wants to face to this challenge by relating objects based on different biological logics. On the other hand, Ontology and enrichment analysis of biological components try to solve this problem based on data reduction and mining. From the well-known projects such as gene ontology and KEGG to newly developed databases are applied to infer biological events reflected from large-scale data. The enrichment analysis softwares are amongst the common computational biology and bioinformatic software. The focus of such software is to find the probability of occurrence of the desired biological features in any arbitrary list of genes/proteins. We introduce Post translational modification Enrichment Integration and Matching Analysis (PEIMAN) software to explore more probable and enriched PTMs on proteins. In this lecture PEIMAN and some other tools are reviewed.