In heimannch/abcnet: Compute scores of an abundant-concordant cytokine network

1. Preparing the input data

1.1. Uploading the scaffold data and filtering it to cells and genes of interest

library(dplyr)
scaffold <- readr::read_tsv("inst/extdata/try_3a.tsv")

The construction of the network is limited to the nodes that we have data for (more about this in the next section). In addition, we can narrow down the network to a set of cells and genes of interest.

cois <- c("B_cells", "T_cells_CD4", "Dendritic_cells")

gois <- c("CANX","CCL5", "IL1B", "FGG", "ITGAL", "ITGB6", "LGALS9", "LAG3", "CXCL9", "IL10", "IFNAR1", "IL10RB", "KIR2DS4", "CD28","BTLA","VTCN1","HLA-B","TNFRSF14", "HLA-A")

It is possible to filter the scaffold based on the selection of cells and genes of interest.

filtered_scaffold <- abcnet::get_scaffold(scaffold, cois, gois)

This filter selects all edges with at least one gene of interest and all edges that connect a cell of interest to a gene of interest. That said, it is plausible that genes outside the gois object are included in the filtered_scaffold, requiring an update in the list of the genes of interest.

gois <- union(filtered_scaffold$From, filtered_scaffold$To) %>% setdiff(cois)

1.2. Creating a random gene expression dataset

#Creating 50 random patients IDs
SampleIDs <-  as.data.frame(do.call(paste0, replicate(5, sample(LETTERS, 100, TRUE), FALSE)))
colnames(SampleIDs) <- "SampleID"

#Creating 2 random groups for the SampleIDs
group <- c("G1", "G2")

SampleIDs <- SampleIDs %>% 
  mutate(Group = sample(group, nrow(.), replace = TRUE))

#Listing genes and cells IDs and creating random expression data
Nodes <- c("BTLA","VTCN1","CANX","CCL5","HLA-B","TNFRSF14", "B_cells")

expr_data <- merge(SampleIDs, Nodes)
expr_data <- expr_data %>% 
  dplyr::mutate(value = rnorm(mean = 10, sd = 5, nrow(.)))

head(expr_data)

In this random dataset, we have 50 SampleIDs, each mapping to a specific group G1 or G2. Each row of this dataset contains the expression value/cell estimate of a gene or cell (y column) for a given Sample ID. This dataframe is in the tidy format required by the functions to compute abundance and concordance scores.

2. Computing the network scores

2.1. Computing the abundance scores

The function abcnet::compute_abundance computes the abundance score for the nodes in a network.

nodes_abundance <- abcnet::compute_abundance(expr_data, node = "y", ids = "SampleID", exprv = "value", group = "Group", cois, gois)

nodes_abundance$Node[1]
nodes_abundance$RatioPerGroup[[1]]

The nodes_abundance dataframe is formatted in a way to make it easier to compute the concordance scores for the edges. To create an object in a format that is easier to export, we can use the get_abundance_table function:

#Organizing the abundance scores for exporting it
nodes_table <- abcnet::get_abundance_table(nodes_abundance)

head(nodes_table)

2.2. Computing the edges scores

To compute the concordance scores of the edges, we just need the scaffold of interest and the nodes scores, as computed by the compute_abundance function.

edges_concordance <- abcnet::compute_concordance(filtered_scaffold, nodes_abundance)

edges_concordance

2.3. Filtering the network based on abundance and concordance thresholds

Especially in the case of networks with hundreds of edges, you may be interested only in nodes and edges with higher values of, respectively, abundance and concordances scores. In this case, you can filter the edges_concordance dataframe based on values of your interest.

filtered_network <- abcnet::filter_network(nodes_table, edges_concordance, abundance = 0.6, concordance = 1)

head(filtered_network)

The filtered_network object contains the edges from the edges_concordance dataframe that meet the following criteria: - both nodes have a abundance score higher than the selected abundance threshold. - edges with concordance score higher than the the selected concordance threshold.

3. Exporting network to file

The object filtered_network can be exported and used in softwares for network visualizations, such as Cytoscape https://cytoscape.org/

readr::write_tsv(filtered_network, paste("edges_", Sys.Date(),".tsv", sep = ""))

heimannch/abcnet documentation built on Jan. 2, 2021, 5:06 p.m.