plot_network: Plot an undirected PPI network using ggraph

In travis-m-blimkie/networker: Reproducible PPI Network Creation and Visualization in R

Plot an undirected PPI network using ggraph

Description

Plot an undirected PPI network using ggraph

Usage

plot_network(
  network,
  fill_column,
  fill_type,
  cat_fill_colours = "Set1",
  layout = "kk",
  legend = FALSE,
  fontfamily = "Helvetica",
  edge_colour_ = "grey40",
  edge_alpha_ = 0.5,
  edge_width_ = 0.5,
  node_size = c(3, 9),
  node_colour = "grey30",
  int_colour = "grey70",
  fc_up_col = "firebrick3",
  fc_down_col = "#188119",
  label = FALSE,
  label_column,
  label_filter = 0,
  label_size = 5,
  label_colour = "black",
  hub_colour = "blue2",
  label_face = "bold",
  label_padding = 0.25,
  min_seg_length = 0.25,
  force_atlas_params = NULL,
  subnet = TRUE,
  seed = 1,
  ...
)

Arguments

network	tidygraph object, output from build_network
fill_column	Tidy-select column for mapping node colour. Designed to handle continuous numeric mappings (either positive/negative only, or both), and categorical mappings, plus a special case for displaying fold changes from, for example, RNA-Seq data. See fill_type for more details on how to set this up.
fill_type	String denoting type of fill mapping to perform for nodes. Options are: "fold_change", "two_sided", "one_sided", or "categorical".
cat_fill_colours	Colour palette to be used when fill_type is set to "categorical." Defaults to "Set1" from RColorBrewer. Will otherwise be passed as the "values" argument in scale_fill_manual().
layout	Layout of nodes in the network. Supports all layouts from ggraph/igraph, as well as "force_atlas" (see Details), or a data frame of x and y coordinates for each node (order matters!).
legend	Should a legend be included? Defaults to FALSE.
fontfamily	Font to use for labels and legend (if present). Defaults to "Helvetica".
edge_colour_	Edge colour, defaults to "grey40"
edge_alpha_	Transparency of edges, defaults to 0.5
edge_width_	Thickness of edges connecting nodes. Defaults to 0.5
node_size	Numeric vector of length two, specifying size range of node sizes (maps to node degree). Default is c(3, 9).
node_colour	Colour (stroke or outline) of all nodes in the network. Defaults to "grey30".
int_colour	Fill colour for non-seed nodes, i.e. interactors. Defaults to "grey70".
fc_up_col	Colour to use for up regulated nodes when fill_type is set to "fold_change". Defaults to "firebrick3".
fc_down_col	Colour to use for down regulated nodes when fill_type is set to "fold_change". Defaults to "#188119".
label	Boolean, whether labels should be added to nodes. Defaults to FALSE.
label_column	Tidy-select column of the network/data to be used in labeling nodes. Recommend setting to gene_name, which contains HGNC symbols mapped from the input Ensembl IDs via biomaRt.
label_filter	Degree filter used to determine which nodes should be labeled. Defaults to 0. This value can be increased to reduce the number of node labels, to prevent the network from being too crowded.
label_size	Size of node labels, defaults to 5.
label_colour	Colour of node labels, defaults to "black"
hub_colour	Colour of node labels for hubs. The top 2% of nodes (based on calculated hub score) are highlighted with this colour, if label = TRUE.
label_face	Font face for node labels, defaults to "bold"
label_padding	Padding around the label, defaults to 0.25 lines.
min_seg_length	Minimum length of lines to be drawn from labels to points. The default specified here is 0.25, half of the normal default value.
force_atlas_params	List of parameters to tweak node positions when using the Force Atlas layout. The following arguments must be supplied: k, gravity, ks, ksmax, and delta. See ?ForceAtlas2::layout.forceatlas2 for details and default values.
subnet	Logical determining if networks produced by extract_subnetwork should be treated as such, or just as a normal network from build_network.
seed	Number used in call to set.seed() to allow for reproducible network generation. Can be changed to get slightly different layouts, and ensure consistent result between runs.
...	Further parameters can be passed on to ggplot2::theme(), e.g. legend.position

Details

Any layout supported by ggraph can be specified here - see ?layout_tbl_graph_igraph for a list of options. Additionally, there is support for the "force_atlas" method, implemented via the ForceAtlas2 package. Finally, you can also supply a data frame containing coordinates for each node. The first and second columns will be used for x and y, respectively. Note that having columns named "x" and "y" in the input network will generate a warning message when supplying custom coordinates.

Since this function returns a standard ggplot object, you can tweak the final appearance using the normal array of ggplot2 function, e.g. labs() and theme() to further customize the final appearance.

The fill_type argument will determine how the node colour is mapped to the desired column. "fold_change" represents a special case, where the fill column is numeric and whose values should be mapped to up (> 0) or down (< 0). "two_sided" and "one_sided" are designed for numeric data that contains either positive and negative values, or only positive/negative values, respectively. "categorical" handles any other non-numeric colour mapping, and uses "Set1" from RColorBrewer.

Node statistics (degree, betweenness, and hub score) are calculated using the respective functions from the tidygraph package.

If plotting a network created from the extract_subnetwork function, the genes belonging to the extracted pathway (i.e. the contents of the "gene_id" column in the enrichment results) will be highlighted, instead of hub nodes. The colour used here can be controlled via the hub_colour argument, and this behaviour can be turned off altogether by setting the subnet argument to FALSE to return an "standard" network. Additionally, one can set the "starters" attribute of the network being plotted to a custom character vector of Ensembl gene IDs, to allow highlighting a custom selection of nodes.

Value

An object of class "gg"

References

See https://github.com/analyxcompany/ForceAtlas2 for details on this method.

See Also

https://github.com/travis-m-blimkie/networker

travis-m-blimkie/networker documentation built on June 3, 2023, 10:17 p.m.