EGAD: ultra-fast functional analysis of gene networks

Ballouz, Sara; Weber, Melanie; Pavlidis, Paul; Gillis, Jesse

doi:10.1093/bioinformatics/btw695

Cited by 73 publications

(88 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Prediction() and performance() function in the R package ROCR were used to calculate AUROCs (Sing et al, 2005). The 277 AUROC values for GO datasets were calculated by the EGAD package (Ballouz et al, 2016) in R. Basically, it utilizes the "guilt by association" principle that genes with shared GO terms are more likely to connected. Thus, networks normalized and inferred by different methods can be evaluated by hiding a subset of genes GO terms and test whether the hidden GO terms could be predicted from the remaining annotations.…”

Section: Network Performance Evaluationmentioning

confidence: 99%

Construction and Optimization of a Large Gene Coexpression Network in Maize Using RNA-Seq Data

et al. 2017

View full text Add to dashboard Cite

ORCID IDs: 0000-0002-6182-800X (J.H.); 0000-0001-6612-3570 (S.V.); 0000-0002-9564-8146 (K.M.M.).With the emergence of massively parallel sequencing, genomewide expression data production has reached an unprecedented level. This abundance of data has greatly facilitated maize research, but may not be amenable to traditional analysis techniques that were optimized for other data types. Using publicly available data, a gene coexpression network (GCN) can be constructed and used for gene function prediction, candidate gene selection, and improving understanding of regulatory pathways. Several GCN studies have been done in maize (Zea mays), mostly using microarray datasets. To build an optimal GCN from plant materials RNA-Seq data, parameters for expression data normalization and network inference were evaluated. A comprehensive evaluation of these two parameters and a ranked aggregation strategy on network performance, using libraries from 1266 maize samples, were conducted. Three normalization methods and 10 inference methods, including six correlation and four mutual information methods, were tested. The three normalization methods had very similar performance. For network inference, correlation methods performed better than mutual information methods at some genes. Increasing sample size also had a positive effect on GCN. Aggregating single networks together resulted in improved performance compared to single networks.Maize (Zea mays) is the most widely produced crop in United States, and U.S. agriculture accounted for 36% of world maize production in 2015 (USDA, 2016). Maize has also been in the center of genetics research for more than 100 years, including McClintock's pioneering work with transposable elements (reviewed by McClintock, 1983;Fedoroff, 2012). Due to recent technological advances in nucleic acid sequencing and the availability of the maize genome sequence (Schnable et al., 2009), maize genomics research has been greatly expedited.RNA-sequencing (RNA-Seq) has become the favored technique for detecting genomewide expression patterns. RNA-Seq has some advantages over microarray analysis of gene expression, including single base-pair resolution, detection of novel transcripts, and the ability to analyze transcript abundance without existing genome information (reviewed by Wang et al., 2009;Han et al., 2015;Conesa et al., 2016). RNA-Seq data provides information about single nucleotide polymorphisms, which facilitates genomewide association studies (Fu et al., 2013;Li et al., 2013a;Lonsdale et al., 2013;Fadista et al., 2014). Because of its widespread adaptability, greater than 5000 Illumina platform Maize RNA-Seq libraries (Fig. 1A) are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database (Leinonen et al., 2010), adding to the body of data that can be used to study the maize genome.The maize genome is large and heterogeneous, and the genome annotation is still far from complete (Cigan et al., 2005;Ficklin and Feltus, 2011). Although recent work has...

show abstract

Section: Network Performance Evaluationmentioning

confidence: 99%

Construction and Optimization of a Large Gene Coexpression Network in Maize Using RNA-Seq Data

et al. 2017

View full text Add to dashboard Cite

show abstract

“…In the second and third approaches, we relied on the guilt-by-association principle to estimate network predictability. In the second method, we used the ‘predictions’ function of EGAD R package 38 . For each gene, this function counts the number of connected genes annotated with an identical GO term and divides this count by the gene’s degree.…”

Section: Methodsmentioning

confidence: 99%

Ranking genome-wide correlation measurements improves microarray and RNA-seq based global and targeted co-expression networks

Liesecke

Daudu

Bernonville

et al. 2018

Preprint

View full text Add to dashboard Cite

10Co-expression networks are essential tools to infer biological associations between gene products and predict gene annotation. Global networks can be analyzed at the transcriptome wide scale or after querying them with a set of guide genes to capture the transcriptional landscape of a given pathway in a process named Pathway Level Correlation (PLC). A critical step in network construction remains the definition of gene co-expression. In the present work, we compared how Pearson Correlation Coefficient (PCC), Spearman Correlation Coefficient (SCC), their respective ranked values (Highest Reciprocal Rank (HRR)), Mutual Information (MI) and Partial Correlations (PC) performed on global networks and PLCs. This evaluation was conducted on the model plant Arabidopsis thaliana using microarray and differently pre-processed RNA-seq datasets. We particularly evaluated how dataset x distance measurement combinations performed in 5 PLCs corresponding to 4 well described plant metabolic pathways (phenylpropanoid, carbohydrate, fatty acid and terpene metabolisms) and the cytokinin signaling pathway. Our present work highlights how PCC ranked with HRR is better suited for global network construction and PLC with microarray and RNA-seq data than other distance methods, especially to cluster genes in partitions similar to biological subpathways. 11 Introduction 12Constructing global gene co-expression networks is a popular approach to highlight transcriptional relationships (edges) 13 between genes (vertices). The 'Guilt-by-Association' (GBA) principle supposes that genes sharing similar functions are 14 preferentially connected and aims at predicting new functions for proteins by determining how their respective encoding 15 genes are co-expressed with others using a reference dataset containing known gene functions such as the Gene Ontology 16 (GO) 1 . Defining edges connecting genes remains a critical step in global co-expression network construction. Expression 17 data (microarray or RNA-seq) are used to construct expression matrices (genes x samples) and to calculate a distance or 18 a similarity for each possible gene pair. The resulting pairwise distance matrix is then thresholded to obtain an adjacency 19 matrix that discriminates relevant edges. Only edges with a distance below (or a similarity above) the set threshold are 20 considered significant and retained for network construction. The procedure is expected to remove non biologically relevant 21 gene associations while retaining the relevant ones and can be assessed with any reference dataset. Alternatively, guide gene sets 22 may be used to extract more human-readable information from large networks in a process named Pathway-Level Correlation 23 (PLC) 2-6 . This approach aims at capturing the best transcriptional associations of a gene set and at highlighting functional gene 24 groups such as known subpathways in this set. There are two types of approaches to determine transcriptional associations 25 of genes: those that are supervised and those that are unsu...

show abstract

“…For our first experiment, we compared metacell networks to bulk in order to capture similarities and differences at the farthest range of the spectrum (see Methods for details on network construction). We first establish that both networks reflect known biology using a guilt-by-association formalism, in which each network is measured for its ability to reconstruct a partially hidden gene list from preferential connectivity within it, outputting an Area under the ROC curve (AUROC) ( Figure 1B, Supplementary Figure 3) 8 . In the aggregate metacells network, the average AUROC for GO slim and KEGG are 0.64 and 0.63 respectively, and similarly the average AUROC of the aggregate bulk RNAseq network is 0.67 for GO and 0.70 for KEGG ( Figure 1C).…”

Section: Mainmentioning

confidence: 99%

Multiscale Co-Expression in the Brain

Harris

Crow

Fischer

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Single-cell RNA-sequencing (scRNAseq) data can reveal co-regulatory relationships between genes that may be hidden in bulk RNAseq due to cell type confounding. Using the primary motor cortex data from the Brain Initiative Cell Census Network (BICCN), we study cell type specific co-expression across 500,000 cells. Surprisingly, we find that the same gene-gene relationships that differentiate cell types are evident at finer and broader scales, suggesting a consistent multiscale regulatory landscape.

show abstract

EGAD: ultra-fast functional analysis of gene networks

Abstract: Supplementary data are available at Bioinformatics online.

Cited by 73 publications

References 16 publications

Construction and Optimization of a Large Gene Coexpression Network in Maize Using RNA-Seq Data

Construction and Optimization of a Large Gene Coexpression Network in Maize Using RNA-Seq Data

Ranking genome-wide correlation measurements improves microarray and RNA-seq based global and targeted co-expression networks

Multiscale Co-Expression in the Brain

Contact Info

Product

Resources

About