#' This function sets up the expression matrix from the metacell object.
#' This function sets up the gene expression matrix for co-expression network analysis.
#'
#' @param seurat_obj A Seurat object
#' @param group_name A string containing a group present in the provided group.by column or in the Seurat Idents.
#' @param group_name A string containing a group present in the provided group.by column or in the Seurat Idents. A character vector can be provided to select multiple groups at a time.
#' @param use_metacells A logical determining if we use the metacells (TRUE) or the full expression matrix (FALSE)
#' @param group.by A string containing the name of a column in the Seurat object with cell groups (clusters, cell types, etc). If NULL (default), scWGCNA uses the Seurat Idents as the group.
#' @param multi.group.by A string containing the name of a column in the Seurat object with groups for consensus WGCNA (dataset, sample, condition, etc)
@@ -190,10 +190,6 @@ SetDatExpr <- function(
modules<-GetModules(seurat_obj,wgcna_name)
assay<-params$metacell_assay
print('n_genes:')
print(length(genes_use))
print(head(genes_use))
# use metacells or whole seurat object?
if(use_metacells){
s_obj<-GetMetacellObject(seurat_obj,wgcna_name)
@@ -201,27 +197,31 @@ SetDatExpr <- function(
s_obj<-seurat_obj
}
# check that group.by is in the Seurat object & in the metacell object:
if(!(group.by%in%colnames(s_obj@meta.data))){
m_cell_message<-""
if(use_metacells){m_cell_message<-"metacell"}
stop(paste0(group.by,' not found in the meta data of the ',m_cell_message,' Seurat object'))
<p>Load the snRNA-seq data and the required libraries:</p>
<p>In this tutorial, we cover how to relate co-expression modules to biological and technical variables. Before starting this tutorial, make sure that you have constructed the co-expression network as in the <ahref="articles/basics_tutorial.html">scWGCNA basics</a>.</p>
<p>First Load the snRNA-seq data and the required libraries:</p>
<p>Here we use the function <code>ModuleTraitCorrelation</code> to correlate selected variables with module eigengenes. This function computes correlations for specified groupings of cells, since we can expect that some variables may be correlated with certain modules in certain cell groups but not in others. There are certain types of variables that can be used for this analysis while others should not be used.</p>
<p><strong>Variables that can be used</strong></p>
<ul>
<li>Numeric variables</li>
<li>Categorical variables with only 2 categories, such as “control” and “condition”.</li>
<li>Categorical variables with a sequential relationship. For example, you may have a “disease stage” category ordered by “healthy”, “stage 1”, “stage 2”, “stage 3”, etc. In this case, you must ensure that the variable is stored as a factor and that the levels are set appropriately.</li>
</ul>
<p><strong>Variables that can not be used</strong></p>
<ul>
<li>Categorical variables with more than two categories that are not sequentially linked. For example, suppose you have a dataset consiting of three strains of transgenic mice and one control. Categorical variables must be converted to numeric before running the correlation, so you will end up with a correlation that is not at all biologically meaningful since there’s not a way to order the three different strains in a way that makes sense as a numeric variable. In this case, you should just set up a pairwise correlation between control and each strain separately. We often have a “Sample ID” variable indicating which cell came from which sample, and this is a variable that does not necessarily make sense to order in any particular way, so a variable like this would not be suitable for module-trait correlation analysis.</li>
<p>For any categorical variables used, this function prints out a warning message to tell the user what order the categories are listed in, just to make sure that it makes sense.</p>
<details><summary>
See warning message
</summary><pre><code>Warning message:
In ModuleTraitCorrelation(seurat_obj, traits = cur_traits, group.by = "cell_type") :
Trait msex is a factor with levels 0, 1. Levels will be converted to numeric IN THIS ORDER for the correlation, is this the expected order?</code></pre>
</details><divclass="section level3">
<h3id="inspecting-the-output">Inspecting the output<aclass="anchor"aria-label="anchor"href="#inspecting-the-output"></a>
</h3>
<p>We can run the function <code>GetModuleTraitCorrelation</code> to retrieve the output of this function.</p>
<p><code>mt_cor</code> is a list containing three items; <code>cor</code> which holds the correlation results, <code>pval</code> which holds the correlation p-values, and <code>fdr</code> which holds the FDR-corrected p-values.</p>
<p>Each of these items is a list where each element is a dataframe for each of the correlation tests that were performed.</p>
<p>We can plot the results of our correlation analysis using the <code>PlotModuleTraitCorrelation</code> function. This function creates a separate heatmap for each of the correlation matrices, and then assembles them into one plot using patchwork.</p>
