broadinstitute/cdsr_plots

cdsrplots

cdsrplots contains standard CDS plots and themes.

Install

library(devtools)
devtools::install_github("broadinstitute/cdsr_plots")

The package can then be loaded by calling

library(cdsrplots)

Functions

• Plotting
  • make_volcano
  • make_gsea_dot
  • make_gsea_bar
• Theme
  • theme_publication
  • scale_fill_publication
  • scale_color_publication

Volcano plot

The make_volcano function is a quick and easy tool to visualize results from differential expression/dependency analyses and can be easily customized.

Example data

As an example we will load the results of a differential expression anlyses comparing Nutlin treated cells to DMSO treated cells.

nutlin <- read_csv("./nutlin.csv")
nutlin %>% head()

## # A tibble: 6 x 3
##   gene   logFC   p_value
##   <chr>  <dbl>     <dbl>
## 1 CDKN1A 1.31  2.82e-166
## 2 MDM2   0.994 5.12e-155
## 3 GDF15  0.977 3.33e-146
## 4 SUGCT  0.708 1.03e-139
## 5 RPS27  0.366 3.80e- 97
## 6 FDXR   0.679 1.04e- 96

Base plot

Make a simple plot by providing the data frame and variable names for effect size and significance.

cdsrplots::make_volcano(nutlin, 'logFC', 'p_value')

Add labels

Provide label_var, a column containing labels for the data points.

• By default rank_by = ‘effect’ and n_labeled = 10, meaning that the 10 left-/right-most points will be labeled, ranked by effect size
• If rank_by = ‘pval’, the n_labeled most significant points will be labeled
• Otherwise, user may specify custom points to label with logical vector label_bool.

cdsrplots::make_volcano(nutlin, 'logFC', 'p_value', label_var = 'gene')

nutlin %>% dplyr::mutate(top_5_sig = rank(p_value) <= 5) %>% 
  cdsrplots::make_volcano('logFC', 'p_value', label_var = 'gene', 
                          label_bool ='top_5_sig',ggrepel_type = 'label')

Add colors

Provide color_var, a categorical column for coloring purposes.

• By default, a grey background/red highlight dual color scheme is used for logical vectors. For all other vector classes, colors are set to categories arbitrarily.
• Otherwise, user may specify custom colors to use with color_values.

nutlin %>% dplyr::mutate(fdr = p.adjust(p_value, method = 'fdr')) %>% 
  cdsrplots::make_volcano('logFC', 'p_value', q_var = 'fdr')

location_colors <- c('left' = '#D95F02', 'right' = '#7570B3', 'bottom' = '#333333')

nutlin %>% 
  dplyr::mutate(location = ifelse(logFC < 0, 'left', 'right')) %>% 
  dplyr::mutate(location = ifelse(p_value > 1e-10, 'bottom', location)) %>% 
  cdsrplots::make_volcano('logFC', 'p_value', color_var = 'location', color_values = location_colors)

Extending functionality with ggplot

Since make_volcano returns a ggplot object, you can make more tweaks using the ggplot syntax.

cdsrplots::make_volcano(nutlin, 'logFC', 'p_value') + cdsrplots::theme_publication()

volcano <- nutlin %>% dplyr::mutate(top_5_sig = rank(p_value) <= 5) %>% 
  cdsrplots::make_volcano('logFC', 'p_value', label_var = 'gene',color_var = 'top_5_sig',
                          label_bool ='top_5_sig',ggrepel_type = 'label')
volcano + aes(size = top_5_sig) + scale_size_manual(values = c(1,3))

Gene set enrichment plots

The make_gsea_dot and make_gsea_bar functions are a quick and easy way to visualize the results of a gene set enrichment analysis. The functions are designed to work with the cdsrgsea package but they are fexible enough to work with other GSEA packages.

Example data

As an example we will look for enriched gene sets from the HALLMARK collection using the hypergeometric test.

gene_sets <- cdsrgsea::load_gene_sets()

nutlin_gsea <- cdsrgsea::run_hyper(nutlin,gene_sets$Hallmark,gene_var = "gene", rank_var = "logFC")
nutlin_gsea %>% head()

## # A tibble: 6 x 8
##   term          p_value p_adjust odds_ratio direction  size overlap_size overlap
##   <chr>           <dbl>    <dbl>      <dbl> <chr>     <int>        <int> <list> 
## 1 HALLMARK_P5… 5.52e-21 2.76e-19      25.8  pos         200           24 <chr […
## 2 HALLMARK_E2… 6.37e-20 3.19e-18      19.8  neg         200           25 <chr […
## 3 HALLMARK_G2… 7.38e-14 1.85e-12      13.2  neg         200           20 <chr […
## 4 HALLMARK_MY… 8.95e- 8 1.49e- 6       7.61 neg         200           14 <chr […
## 5 HALLMARK_AP… 3.33e- 5 8.32e- 4       6.70 pos         161            9 <chr […
## 6 HALLMARK_TN… 1.81e- 4 3.02e- 3       5.28 pos         200            9 <chr […

Base plot

Make a simple bar plot by providing the data frame returned by cdsrgsea.

cdsrplots::make_gsea_bar(nutlin_gsea)

The default variable names match the variable names returned by cdsrgsea.

• enrich_var specifies the column containing the enrichment values. By default enrich_var = ‘odds_ratio’ for hypergeometric and ‘NES’ for GSEA.
• size_var specifies the column containing the sizes. By default size_var = ‘overlap_size’ for hypergeometric and ‘size’ for GSEA.
• p_var specifies the column contianing the significance values. By default p_var = ‘p_value’

Variable names can changed to work with other GSEA packages or to customize the plots. For example we can setp_var to ‘p_adjust’ instead of ‘p_value’.

cdsrplots::make_gsea_dot(nutlin_gsea,p_var = 'p_adjust')

Single direction

The direction parameter dir can be set to ‘pos’ to only shown positive terms or ‘neg’ to only show negative terms

cdsrplots::make_gsea_dot(nutlin_gsea,dir = "pos")

Change coloring

The color_by argument sets how the plot is colored. There are three options

• ‘pval’ - colors by significance
• ‘dir’ - colors by the direction of the enrichment
• ‘enrich’ - colors by enrichment

cdsrplots::make_gsea_bar(nutlin_gsea, color_by = "dir")

Change x-axis

The x_by argument sets which variable is plotted on the x-axis. There are two options

• ‘pval’ - the significance is plotted on the x-axis
• ‘enrich’ - the enrichment is plotted on the x-axis

cdsrplots::make_gsea_bar(nutlin_gsea, color_by = "enrich", x_by = "pval")

Change y-axis

There are a number of parameters that modify the y-axis

• n_shown sets the number of terms which are shown.
• sig_only determines whether only significant terms are shown.

cdsrplots::make_gsea_dot(nutlin_gsea,sig_only = T)