In UCLouvain-CBIO/depmap: Cancer Dependency Map Data Package

BiocStyle::markdown()

This vignette illustrates use cases and visualizations of the data found in the depmap package. See the depmap vignette for details about the datasets.

Introduction

The depmap package aims to provide a reproducible research framework to cancer dependency data described by Tsherniak, Aviad, et al. "Defining a cancer dependency map." Cell 170.3 (2017): 564-576.. The data found in the depmap package has been formatted to facilitate the use of common R packages such as dplyr and ggplot2. We hope that this package will allow researchers to more easily mine, explore and visually illustrate dependency data taken from the Depmap cancer genomic dependency study.

Use cases

Perhaps the most interesting datasets found within the depmap package are those that relate to the cancer gene dependency score, such as rnai and crispr. These datasets contain a score expressing how vital a particular gene is in terms of how lethal the knockout/knockdown of that gene is on a target cell line. For example, a highly negative dependency score implies that a cell line is highly dependent on that gene.

Load necessary libaries.

library("dplyr")
library("ggplot2")
library("viridis")
library("tibble")
library("gridExtra")
library("stringr")
library("depmap")
library("ExperimentHub")

Load the rnai, crispr and copyNumber datasets for visualization. Note: the datasets listed below are from the 19Q3 release. Newer datasets, such as those from the 20Q1 release are available.

## create ExperimentHub query object
eh <- ExperimentHub()
query(eh, "depmap")
rnai <- eh[["EH3080"]]
mutationCalls <- eh[["EH3085"]]
metadata <- eh[["EH3086"]]
TPM <- eh[["EH3084"]]
copyNumber <- eh[["EH3082"]]
# crispr <- eh[["EH3081"]]
# drug_sensitivity <- eh[["EH3087"]]

By importing the depmap data into the R environment, the data can be mined more effectively. For example, if one interested researching soft tissue sarcomas and wanted to search all such cancer cell lines for the gene with the greatest dependency, it is possible to accomplish this task by using data manipulation and visualization tools dplyr and ggplot2. Below, the rnai dataset is selected for cell lines with "SOFT_TISSUE" in the CCLE name, and displaying a list of the highest dependency scores.

## list of dependency scores
rnai |>
    dplyr::select(cell_line, gene_name, dependency) |>
    dplyr::filter(stringr::str_detect(cell_line, "SOFT_TISSUE")) |>
    dplyr::arrange(dependency) |>
    head(10)

As the gene RPL14 appears several times in the top dependencies scores, it may make an interesting candidate target. Below, a plot of the rnai data is displayed as a histogram showing the distribution of dependency scores for gene RPL14.

## Basic histogram
rnai |>
    dplyr::select(gene, gene_name, dependency) |>
    dplyr::filter(gene_name == "RPL14") |>
    ggplot(aes(x = dependency)) +
    geom_histogram() +
    geom_vline(xintercept = mean(rnai$dependency, na.rm = TRUE),
               linetype = "dotted", color = "red") +
    ggtitle("Histogram of dependency scores for gene RPL14")

A more complex plot of the rnai data, as shown below involves plotting the distribution of dependency scores for gene RPL14 for each major type of cancer, while highlighting the nature of mutations of this gene in such cancer cell lines (e.g. if such are COSMIC hotspots, damaging, etc.). Notice that the plot above reflects the same overall distribution in two dimensions.

meta_rnai <- metadata |>
    dplyr::select(depmap_id, lineage) |>
    dplyr::full_join(rnai, by = "depmap_id") |>
    dplyr::filter(gene_name == "RPL14") |>
    dplyr::full_join((mutationCalls |>
                      dplyr::select(depmap_id, entrez_id,
                                    is_cosmic_hotspot, var_annotation)),
                     by = c("depmap_id", "entrez_id"))

p1 <- meta_rnai |>
      ggplot(aes(x = dependency, y = lineage)) +
      geom_point(alpha = 0.4, size = 0.5) +
      geom_point(data = subset(
         meta_rnai, var_annotation == "damaging"), color = "red") +
      geom_point(data = subset(
         meta_rnai, var_annotation == "other non-conserving"), color = "blue") +
      geom_point(data = subset(
         meta_rnai, var_annotation == "other conserving"), color = "cyan") +
      geom_point(data = subset(
         meta_rnai, is_cosmic_hotspot == TRUE), color = "orange") +
      geom_vline(xintercept=mean(meta_rnai$dependency, na.rm = TRUE),
                 linetype = "dotted", color = "red") +
      ggtitle("Scatterplot of dependency scores for gene RPL14 by lineage")

p1

Below is a boxplot displaying expression values for gene RPL14 by lineage:

metadata |>
    dplyr::select(depmap_id, lineage) |>
    dplyr::full_join(TPM, by = "depmap_id") |>
    dplyr::filter(gene_name == "RPL14") |>
    ggplot(aes(x = lineage, y = expression, fill = lineage)) +
    geom_boxplot(outlier.alpha = 0.1) +
    ggtitle("Boxplot of expression values for gene RPL14 by lineage") +
    theme(axis.text.x = element_text(angle = 45, hjust=1)) +
    theme(legend.position = "none")

High dependency, high expression genes are more likely to interesting research targets. Below is a plot of expression vs rnai gene dependency for Rhabdomyosarcoma Sarcoma:

## expression vs rnai gene dependency for Rhabdomyosarcoma Sarcoma
sarcoma <- metadata |>
    dplyr::select(depmap_id, cell_line,
                  primary_disease, subtype_disease) |>
    dplyr::filter(primary_disease == "Sarcoma",
                  subtype_disease == "Rhabdomyosarcoma")

rnai_sub <- rnai |>
    dplyr::select(depmap_id, gene, gene_name, dependency)
tpm_sub <- TPM |>
    dplyr::select(depmap_id, gene, gene_name, expression)

sarcoma_dep <- sarcoma |>
    dplyr::left_join(rnai_sub, by = "depmap_id") |>
    dplyr::select(-cell_line, -primary_disease,
                  -subtype_disease, -gene_name)

sarcoma_exp <- sarcoma |>
    dplyr::left_join(tpm_sub, by = "depmap_id")

sarcoma_dat_exp <- dplyr::full_join(sarcoma_dep, sarcoma_exp,
                                    by = c("depmap_id", "gene")) |>
    dplyr::filter(!is.na(expression))

p2 <- ggplot(data = sarcoma_dat_exp, aes(x = dependency, y = expression)) +
    geom_point(alpha = 0.4, size = 0.5) +
    geom_vline(xintercept=mean(sarcoma_dat_exp$dependency, na.rm = TRUE),
               linetype = "dotted", color = "red") +
    geom_hline(yintercept=mean(sarcoma_dat_exp$expression, na.rm = TRUE),
               linetype = "dotted", color = "red") +
    ggtitle("Scatterplot of rnai dependency vs expression values for gene")

p2 + theme(axis.text.x = element_text(angle = 45))

A selection of the genes shown above with the lowest depenency scores, also displaying gene expression in TPM in the last column.

sarcoma_dat_exp |>
    dplyr::select(cell_line, gene_name, dependency, expression) |>
    dplyr::arrange(dependency) |>
    head(10)

Below is a boxplot displaying log genomic copy number for gene RPL14 by lineage:

metadata |>
    dplyr::select(depmap_id, lineage) |>
    dplyr::full_join(copyNumber, by = "depmap_id") |>
    dplyr::filter(gene_name == "RPL14") |>
    ggplot(aes(x = lineage, y = log_copy_number, fill = lineage)) +
    geom_boxplot(outlier.alpha = 0.1) +
    ggtitle("Boxplot of log copy number for gene RPL14 by lineage") +
    theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
    theme(legend.position = "none")