Multivariate analysis is an analysis of more than one statistical outcome variable at a time. In meta-omics studies, the data are high dimensional, i.e. each feature (protein, gene, metabolite) is a dimension across the samples.
This app is a wrapper of several methods of multivariate analysis for your high dimensional data. The following functions are included:
- Unsupervised approaches, including clustering, PCA and t-SNE
- Hierarchical clustering in combination with heatmap helps visualize the high dimensional data using heat colors.
- PCA and t-SNE are both for dimensionality reduction and visualization. PCA is a linear dimension reduction method, while t-SNE is non-linear.
- Supervised approach, PLS-DA.
- PLS-DA uses your meta table as grouping information to supervise the study. A valid PLS-DA model finds out the features that are responsible for the between-group differences.