3mcor

	Metabolome	Microbiome
Correlation method	The method of correlation.	The method of correlation.

If eliminate confounding effect?

Global correlation

This pipeline is the highest-level analysis aiming at the evaluation of overall correlation between two omics data sets.

Use Phonetype in correlation?

*Please be consistent with the category of the phenotype

Method of the global correlation

CCA Canonical correlation analysis (CCA) projects multidimensional data X and Y into variables X' and Y' using linear transformation and then take the correlation coefficient of X' and Y' as the correlation of X and Y.	CIA Coinertia analysis (CIA) is a two-table ordination method used to measure the consistency between two datasets and can be considered as a PCA of the joint covariances of X and Y.	PA Procrustes analysis (PA) compares the consistency of two data sets by analyzing the shape distributions of scatter plots derived from other dimension reduction and projection methods.
O2PLS Two-way orthogonal partial least squares (O2PLS) breaks the data into five parts: the joint variance of X and Y, the part of X Unique, the part of Y Unique, the residual part of X, and the residual part of Y. The Pearson correlation between the joint variance of X and Y is taken as the global correlation of the two data sets.	Mantel Test The Mantel Test calculates the correlation of two unfolded variables of the distance matrix between data sets X and Y.	sPLS Sparse partial least squares (sPLS) is achieved with a Lasso penalization of the PLS loading vectors when computing the Singular Value Decomposition. The Pearson correlation of the first principal components derived from the Predictor matrixes is taken as the global correlation of two data sets.

The method of global correlation. "CCA" is the default selection and the multi-selection is allowed.

**The number of samples must exceed the number of variables for CCA.

Hierarchical correlation

This pipeline is summarized and simplified in 3 steps. First, users select a dimension reduction method to get the metabolic (Meta) and microbial (Mic) modules. Second, phenotype (if exist) will be used for module filtering. Third, the importance of variables in each module are ranked by their within-module connectivity (for WGCNA) or their loading value of first principal component (for PCA/PCoA).

Method of dimension reduction

Method of dimension reduction.The modules of omics data will be generated.

Use Phonetype in correlation?

*Please be consistent with the category of the phenotype

Method of cluster-wide correlation

Pairwise correlation

This pipeline is the lowest-level analysis in which pairwise correlation between metabolic and microbial variables will be evaluated. Besides pairwise correlation analysis, network and topological analysis were incorporated hoping to highlight core nodes or links among the massive correlated pairs.

Use Phonetype in correlation?

*Please be consistent with the category of the phenotype

Method of pairwise correlation

The method of correlation.

Threshold values in the network plot

r value
p value
Centrality The parameter that calculates the importance of a network node.

CORRELATION