A powerful framework for an integrative study with heterogeneous omics data: from univariate statistics to multi-block analysis

Abstract

The high-throughput data generated by new biotechnologies used in biological studies require specific and adapted statistical treatments. In this work, we propose a novel and powerful framework to manage and analyse multi-omics heterogeneous data to carry out an integrative analysis. We illustrate it using the package mixOmics for the R software as it specifically addresses data integration issues. Our work also aims at confronting the most recent functionalities of mixOmics to real data sets because, even if multi-block integrative methodologies exist, they still have to be used to enlarge our know-how and to provide an operational framework to biologists. Natural populations of the model plantArabidopsis thalianaare employed in this work but the framework proposed is not limited to this plant and can be deployed whatever the organisms of interest and the biological question. Four omics data sets (phenomics, metabolomics, cell wall proteomics and transcriptomics) have been collected, analysed and integrated in order to study the cell wall plasticity of plants exposed to sub-optimal temperature growth conditions. The methodologies presented start from basic univariate statistics and lead to multi-block integration analysis, and we highlight the fact that each method is associated to one biological issue. Using this powerful framework led us to novel biological conclusions that could not have been reached using standard statistical approaches.

An Integrative Study Showing the Adaptation to Sub-Optimal Growth Conditions of Natural Populations of Arabidopsis thaliana: A Focus on Cell Wall Changes

“…Even if the principal component analysis (PCA) is an unsupervised method, it clearly highlighted groups of samples and allowed evaluating the good reproducibility of a given experiment [ 40 ], when performed on quantified CWPs ( Figure 4 ). The analysis of the two organs showed a growth temperature-specific response of each population at the CW proteome level.…”

“…Thereafter, we refer to each omics dataset (phenomics, metabolomics, proteomics and transcriptomics) as a block and we have used a previously described framework for the integration of the omics data [ 40 ].…”

“…Rosettes were collected just before bolting (stage 5.10 [ 41 ]) from four- or six-week-old plants grown at 22 or 15 °C, respectively. Floral stems were also collected at the same stage of development (first flower, stage 6 [ 40 ]): 6 weeks for Col; 7 weeks for Grip and Roch; and 8 weeks for Hosp and Hern. They were collected two weeks later for all the plants grown at 15 °C.…”

“…Data analysis and methodology were achieved as described [ 40 ]. We have performed multi-block supervised analyses with the sparse version Multi-Block Partial Least Squares Discriminant Analysis (MB-PLS-DA) implemented in the mixOmics R package version 6.1.1.…”

“…To tackle the complex mechanisms of adaptation to sub-optimal temperatures, a systems biology approach combining climate parameters, phenomics, metabolomics, proteomics and transcriptomics has been developed. The statistical integrative analysis of the various heterogeneous omic datasets using the mixOmics package [ 39 , 40 ] allowed identifying new genes/proteins possibly involved in the phenotypic plasticity responses of the four Pyrenean populations and the Columbia ecotype (Col) as a reference to temperature changes, with a specific focus on the CW.…”