In chiasinL/STRINGutils: Additional utilities for STRINGdb

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

suppressPackageStartupMessages(c(
  library(STRINGutils),
  library(STRINGdb),
  library(magrittr)
))

Introduction

STRINGdb R package provides a convenient way to access the popular STRING database (https://string-db.org) from R. STRINGutils is built on top of STRINGdb and provides convenient functions to get SVG file of STRING network and highlight features of interest.

To prepare data necessary for STRINGutils, we first run code from STRINGdb vignette.

Briefly, we map the analyzed data of a microarray study to a human STRING database to get STRING identifiers. Please see STRINGdb vignette for more information.

string_db <- STRINGdb$new(version = "11", species = 9606,
                          score_threshold = 200,
                          input_directory = "")

data("diff_exp_example1")

example1_mapped <- string_db$map(diff_exp_example1, "gene", removeUnmappedRows = TRUE)

hits <- example1_mapped$STRING_id[1:200]

Get SVG of STRING network

Once we have string_db, an instantiated STRINGdb reference class with associated information from STRING database, we can run the get SVG function. If file is specified, the SVG will be saved.

xml <- get_svg(string_db, hits)

xml <- get_svg(string_db, hits, file = "my_network.svg")

Get full or subnetwork for proteins of interest

If we have a list of proteins of interest, the full or sub- STRING network of the proteins can be obtained. These proteins will be highlighted by the colors of choice.

For illustration purposes, here we will pick the top 10 proteins.

hits_filt <- example1_mapped %>%
  dplyr::slice_min(pvalue, n = 10, with_ties = FALSE) %>%
  dplyr::filter(!is.na(STRING_id)) # <- filter out proteins with no mapped STRING IDs

Let's take a look at the proteins of interest. It is good to know that none of the top 10 proteins are missing STRING IDs.

knitr::kable(hits_filt, align = "ccc")

Next, we need to make a color vector of the same length as our proteins of interest to specify what colors to highlight our proteins with.

colors_vec <- rep("rgb(101,226,11)", nrow(hits_filt))
names(colors_vec) <- hits_filt$gene 

Let's take a look at the color vector.

colors_vec

Note: It is important to double check to make sure that the proteins of interest are matched up to their intended colors.

Now, we can go ahead and get the network. The function saves the network as a SVG file in the current directory as "features_of_int.svg".

For full network:

plot_features(example1_mapped, colors_vec, string_db, entire = TRUE)

The SVG file can be viewed with the magick::image_read_svg() function in RStudio viewer.

magick::image_read_svg("features_of_int.svg")

The default number of proteins obtained from the network is set to 200. This number can be changed with n_hits. The network view mode for STRING network can also be defined with network_flavor. More information about this can be found at STRING's website.

For example, to increase the number of proteins in the network to 500 and use the network view mode of confidence instead of the default evidence:

plot_features(example1_mapped, colors_vec, string_db, n_hits = 500, entire = TRUE,
              network_flavor = "confidence")

magick::image_read_svg("features_of_int.svg")

For subnetwork:

plot_features(example1_mapped, colors_vec, string_db)

magick::image_read_svg("features_of_int.svg")

chiasinL/STRINGutils documentation built on Dec. 19, 2021, 3:56 p.m.