What is Correlation analysis?

Correlation analysis is a statistical test that lets you rank features by their correlation with numeric attributes using Pearson (linear), Spearman (rank), or Kendall (tau) correlation. 

Running Correlation analysis

We recommend normalizing you data prior to running Correlation analysis, but it can be invoked on any counts data node. 

• Click the counts data node
• Click the Statistics section in the toolbox
• Click Correlation 
• Choose the method to use for correlation analysis (Figure 1) 

Feature many-to-one correlation

Only numeric factors are available. When multiple factors are added, the correlation analysis will perform each factor with a feature in the data node independently.

• Select the factors and interactions to include in the statistical test (Figure 2)

• Click Next 
• It is optional to apply a lowest coverage filter or configure the advanced settings
• Click Finish to run

Correlation analysis produces a Correlation data node; double-click to open the task report (Figure 3) which is similar to the ANOVA and GSA task reports and includes a table with features on rows and statistical results on columns.

Each numeric attribute includes p-value, adjusted p-value columns (FDR step up and/or Storey q-value if included), and a partial correlation value. Each interaction will have p-value and adjusted p-value columns (FDR step up and/or Storey q-value if included).

Each feature includes  chromosome view,  dot plot,  correlation plot, and extra details  buttons in the View column. 

Correlation analysis advanced options

Low value filter

Low-value filter allows you to specify criteria to exclude features that do not meet the requirements for the calculation. If there is a filter feature task performed in the upstream analysis, the default of this filter is set to None, otherwise, the default is Lowest average coverage is set to 1.

Lowest average coverage: the computation will exclude a feature if its geometric mean across all samples is below the specified value

Lowest maximum coverage: the computation will exclude a feature if its maximum across all samples is below the specified value

Minimum coverage: the computation will exclude a feature if its sum across all samples is below the specified value

None: include all features in the computation

Multiple test correction

Multiple test correction can be performed on the p-values of each comparison, with FDR step-up being the default. If you check the Storey q-value, an extra column with q-values will be added to the report.

Use only reliable estimation results

There are situations when a model estimation procedure does not fail outright but still encounters some difficulties. In this case, it can even generate p-value and fold change on the comparisons, but they are not reliable, i.e. they can be misleading. Therefore, the default of Use only reliable estimation results is set Yes.

Correlation type

Sets the type of correlation used to calculate the correlation coefficient and p-value. Options are Pearson (linear), Spearman (rank), Kendall (tau). Default is Pearson (linear). 

Correlation across assays

Correlation across assays should be used to perform correlation analysis across different modalities (e.g. ATAC-Seq enriched regions vs. RNA-Seq expression) for multiomics data analysis. This option will only be available when multiple types of data are available in the project. 

• Select the data node to be compared to the node that the task has been invoked from using the Select data node button
• Modify any parameters (Figure 4)
• Click Finish

Correlation across assays produces a Correlation pair list data node; double-click to open the list.