ATTED-II in 2018: A Plant Coexpression Database Based on Investigation of the Statistical Property of the Mutual Rank Index

Abstract

ATTED-II (http://atted.jp) is a coexpression database for plant species to aid in the discovery of relationships of unknown genes within a species. As an advanced coexpression analysis method, multispecies comparisons have the potential to detect alterations in gene relationships within an evolutionary context. However, determining the validity of comparative coexpression studies is difficult without quantitative assessments of the quality of coexpression data. ATTED-II (version 9) provides 16 coexpression platforms for nine plant species, including seven species supported by both microarray- and RNA sequencing (RNAseq)-based coexpression data. Two independent sources of coexpression data enable the assessment of the reproducibility of coexpression. The latest coexpression data for Arabidopsis (Ath-m.c7-1 and Ath-r.c3-0) showed the highest reproducibility (Jaccard coefficient = 0.13) among previous coexpression data in ATTED-II. We also investigated the statistical basis of the mutual rank (MR) index as a coexpression measure by bootstrap sampling of experimental units. We found that the error distribution of the logit-transformed MR index showed normality with equal variances for each coexpression platform. Because the MR error was strongly correlated with the number of samples for the coexpression data, typical confidence intervals for the MR index can be estimated for any coexpression platform. These new, high-quality coexpression data can be analyzed with any tool in ATTED-II and combined with external resources to obtain insight into plant biology.

Identification of gene expression logical invariants in Arabidopsis

“…This study identified Pathway Level Co‐expression using a set of guide genes, and 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. In another study, a co‐expression database for plant species ATTED‐II (http://atted.jp) was published to aid in the discovery of relationships of unknown genes within a species (Obayashi, Aoki, Tadaka, Kagaya, & Kinoshita, ). ATTED‐II (version 9) provides 16 co‐expression platforms for nine plant species, including seven species supported by both microarray‐ and RNA sequencing (RNAseq)‐based co‐expression data (Obayashi et al., ).…”

“…In another study, a co‐expression database for plant species ATTED‐II (http://atted.jp) was published to aid in the discovery of relationships of unknown genes within a species (Obayashi, Aoki, Tadaka, Kagaya, & Kinoshita, ). ATTED‐II (version 9) provides 16 co‐expression platforms for nine plant species, including seven species supported by both microarray‐ and RNA sequencing (RNAseq)‐based co‐expression data (Obayashi et al., ). Similarly, co‐expression networks have been a popular tool in the literature to understand regulatory pathways in Arabidopsis (He & Maslov, ; Van Bel & Coppens, ; Zheng et al., ).…”

Arabidopsis bioinformatics resources: The current state, challenges, and priorities for the future

“…Arabidopsis should be the model to show other organisms what is possible and suggest ways to do it. A key area of future effort is to move analyses done in Arabidopsis into other species, e.g., as ATTED [a coexpression database (Obayashi, Aoki, Tadaka, Kagaya, & Kinoshita, )], PMR [metabolomics and transcriptomic coexpression database (Hur et al., )] and BAR http://bar.utoronto.ca/ are doing. Outreach to other plant groups to convey vital lessons learned would serve to increase interactions between the various populations and allow leveraging of experience in the Arabidopsis model system for additional important species.…”

Computational aspects underlying genome to phenome analysis in plants

“…The only consideration is that expression data sets should represent a range of different conditions and not favor certain perturbations over others. This could be done either by hand, for example in CSBDB.DB (Steinhauser et al ., ), or automatically, for example in ATTED‐II (Obayashi et al ., ).…”