The breast cancer data by Krug et al., 2020 used for this tutorial is already preprocessed and only includes samples with tumor purity >0.5 and known ER status. Metabolite data was sampled randomly to generate distributions similar to those reported e.g. in (Terunuma et al., 2014). Krug et al., 2020 published data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC). The dataset contains observations from:
Load data
First you load the preprocessed data. This data is included in the package and does not need to be manually loaded but can be directly accessed once library(molnet) is called.
data("mrna_data")
data("protein_data")
data("phosphoprotein_data")
data("metabolite_data")
data("metabolite_protein_interaction")
Transform to required input format
Then you can use formatting functions to bring the data into the required input format:
- make_layer() # Creates individual molecular layers from raw data and unique identifiers
- make_connection() # Specifiy connection between two individual layers
- make_drug_target() # Formats drug target interactions
In the beginning individual layer objects are created using . The user supplies data stratified over two groups and identifiers as shown in the code chunk below. The identifiers need to be in the same order as the components in the data. Additionally, the layer is named by the name argument. All individual layers are passed in a single list. Naming of the columns in the respective identifier data frames need to be consistent i.e. if the mRNA layer and the protein layers should be connected both need to contain a column with the same identifier name. In this example mRNA and Proteins both contain gene_name. The data frame drug_gene_interactions should contain identifiers with consistent column names as well as drug names and IDs.
head(protein_data$group1$identifiers)
#> refSeq gene_name STRING_id
#> 1 NP_001254479.2 TTN 9606.ENSP00000467141
#> 2 NP_001243779.1 TTN 9606.ENSP00000467141
#> 3 NP_596870.2 TTN 9606.ENSP00000467141
#> 4 NP_569827.2 MADD 9606.ENSP00000310933
#> 5 NP_036555.1 RPL13A 9606.ENSP00000375730
#> 6 NP_001122305.1 ZBTB4 9606.ENSP00000307858
head(mrna_data$group1$identifiers)
#> gene_name
#> 1 A1BG
#> 2 A1BG-AS1
#> 3 A1CF
#> 4 A2M
#> 5 A2ML1
#> 6 A2MP1
For computational reasons we use a subset of 100 genes for this vignette. Below we create individual layers.
number_of_genes <- 100 # set for subsetting
# Create individual layers
mrna_layer <- make_layer(name="mrna",
data_group1=mrna_data$group1$data[,1:number_of_genes],
data_group2=mrna_data$group2$data[,1:number_of_genes],
identifier_group1=data.frame(gene_name=mrna_data$group1$identifiers[1:number_of_genes,]),
identifier_group2=data.frame(gene_name=mrna_data$group2$identifiers[1:number_of_genes,])
)
protein_layer <- make_layer(name="protein",
data_group1=protein_data$group1$data[,1:number_of_genes],
data_group2=protein_data$group2$data[,1:number_of_genes],
identifier_group1=protein_data$group1$identifiers[1:number_of_genes,],
identifier_group2=protein_data$group2$identifiers[1:number_of_genes,]
)
The pipeline requires a list containing all individual layers. Below we are creating the list.
layers <- list(
mrna_layer,
protein_layer,
phosphoprotein_layer,
metabolite_layer
)
Then you need to supply the inter-layer connections. From and to need to match with a name in the previously created layers by . The established connection will result in an undirected graph. It is possible to create connections based on mutual identifiers present in both connected layers or based on an interaction table supplied that you need to supply. In case connections are matched on identifiers connect_on specifies the shared identifier and the edge weight is passed to weight. The default weight is 1. In case connections are established using an interaction table it has to be passed to connect_on and weight specifies the name of the column in the interaction table which is used as edge weight.
inter_layer_connections = list(
make_connection(from = 'mrna', to = 'protein', connect_on = 'gene_name', weight = 1),
make_connection(from = 'protein', to = 'phosphoprotein', connect_on = 'gene_name', weight = 1),
make_connection(from = 'protein', to = 'metabolite', connect_on = metabolite_protein_interaction, weight = 'combined_score')
)
For running the pipeline drug-target interactions are required. The the function generates the required format. The argument interaction_table maps drugs to targets. The argument match_on specifies the column of the interaction_table used to match drugs to targets.
drug_target_interaction <- make_drug_target(target_molecules='protein',
interaction_table=drug_gene_interactions,
match_on='gene_name')
When the input layers, connection and drug targets are created they are checked automatically for validity. The function below checks for a variety of possible input formatting errors.
return_errors(check_input(layers = layers, inter_layer_connections = inter_layer_connections, drug_target_interaction = drug_target_interaction))
[image created with biorender.com]
[image created with biorender.com]