Differential Expression Analysis

Differential expression analysis aims to discover quantitative changes in gene-expression levels between defined experimental groups. In SPATA these experimental groups are defined inside the feature data. More precisely: Every discrete variable that is part of the spata-object’s feature data assigns every sample’s barcode-spot to such an experimental group. This includes all spata-intern generated groups such as segmentation and clustering as well as any other discrete feature of your own extracted analysis that has been added via addFeature(). See Extract, add & join data on how to do that with ease.

Differential expression analysis

The following tutorial steps will guide you through the spata-intern differential gene expression functions using a previously created segmentation as the experimental groups.

# load packages
library(SPATA)
library(magrittr)
library(ggplot2)

# load object
spata_obj <- loadSpataObject(input_path = "data/spata-obj-de-analysis-example.RDS")

plotSegmentation(spata_obj)
Figure 1. Spatial segmentation of two areas of interest.

Figure 1. Spatial segmentation of two areas of interest.

1. Find differentially expressed genes

As mentioned in previous chapters SPATA makes use of some well written functions of the Seurat-package which currently define the gold-standard of statistical scRNA-seq analysis. The function findDE()is a wrapper around Seurat::FindAllMarkers() and lets you denote the experimental groups across which you want to analyze diferentially expressed genes. You can do that via the across-argument which we are going to denote as segment as this is the feature we are interested in for now. (If you are only interested in a subset of groups the specified variable contains use the across_subset-argument to filter for those).

de_results <-
  findDE(object = spata_obj,
         across = "segment", # denotes experimental group belonging of interest
         across_subset = c("oxid-phosph-high", "hypoxic"), # subsets the groups
         method_de = "wilcox",
         p_val_adj = 0.05)

# output
de_results

2. Filter according to your needs

In order to post-process the resulting de-data.frame use filterDE(). The data.frame will be sliced in order that for every cluster n-genes are filtered depending on the input of n_highest_FC and n_lowest_pvalue.

filtered_de_results <-
  filterDE(de_df = de_results,
           n_highest_FC = 100, # keep the 100 genes with the highest log fold change
           n_lowest_pvalue = 50, # from these 100 genes keep the 50 genes with the lowest p-value
           return = "data.frame")

filtered_de_results

3. Plot your results

The gold standard of differentially gene expression visualization is the classical heatmap. plotDeHeatmap() takes your DE-results and plots the respective heatmap by extracting the genes and barcode-spots of interest. Via additional computation the heatmap is segmented into clear and aesthetically pleasing rectangulars. Additional arguments to pheatmap::pheatmap() can be specified via ....

heatmap <-
  plotDeHeatmap(object = spata_obj,
                de_df = filtered_de_results, # specify the data
                across = "segment",  # specify the feature
                across_subset = unique(filtered_de_results$cluster),
                hm_colors = viridis::inferno(n = 15), # provide your color spectrum of choice
                breaks = c( c(-5.5, -0.6), seq(-0.5,1.8, length.out = 10), c(1.9, 3)),
                show_rownames = FALSE)

heatmap
Figure 2. Heatmap visualizing the differentially expressed genes

Figure 2. Heatmap visualizing the differentially expressed genes

In order to visualize only a subset of genes make use of plotDistributionAcross() which plots the distribution of specific variables across specific subgroups.

genes_of_interest <-
    filterDE(de_df = de_results,
             n_highest_FC = 10,
             n_lowest_pvalue = 10,
             return = "vector") # obtain only a vector of genes as input for 'variables'

plotDistributionAcross(spata_obj,
                       variables = genes_of_interest,
                       across = "segment",
                       across_subset = c("oxid-phosph-high", "hypoxic"),
                       plot_type = "violin") +
  theme(axis.text.x = element_blank(),
        legend.position = "top")
Figure 2.2) Violinplot visualizing the differentially expressed genes

Figure 2.2) Violinplot visualizing the differentially expressed genes

Another way to visualize your results would be to deploy plotSurfaceComparison().

hypoxic_high_genes <- genes_of_interest <-
    filterDE(de_df = de_results,
             n_highest_FC = 10,
             n_lowest_pvalue = 10,
             across_subset = "hypoxic",
             return = "vector")

plotSurfaceComparison(spata_obj,
                      variables = hypoxic_high_genes,
                      smooth = TRUE,
                      pt_size = 1)
Figure 2.3) Surfaceplots visualizing the differentially expressed genes

Figure 2.3) Surfaceplots visualizing the differentially expressed genes