examples/save_data_objects.R

In taylorpourtaheri/nr: Node prioritization optimizer

# save STRING db object
string_db <- STRINGdb::STRINGdb$new(version="11",
                                    species=9606,
                                    score_threshold = 950)
saveRDS(string_db, 'data/string_db_v11.RDS')

# save PPI object
ppi <- string_db$get_graph()
saveRDS(ppi, 'data/string_ppi_v11.RDS')

# save gene ID/symbol xref
ensembl <- biomaRt::useMart("ensembl", dataset = "hsapiens_gene_ensembl")
xref <- biomaRt::getBM(attributes = c('ensembl_peptide_id', 'ensembl_gene_id',
                                      'hgnc_symbol', 'external_gene_name',
                                      'description'),
                       mart = ensembl)
saveRDS(xref, 'data/biomart_xreference.RDS')

xref_subset <- data.frame('symbol' = unique(xref$hgnc_symbol))
xref_subset <- string_db$map(xref_subset, "symbol", removeUnmappedRows=T, quiet=T)
xref_subset <- dplyr::filter(xref_subset, STRING_id %in% igraph::V(ppi)$name)
saveRDS(xref_subset, 'data/biomart_xreference_ppi_genes.RDS')

# save similarity matrix (jaccard is the default method)
sim <- igraph::similarity(ppi)
saveRDS(sim, 'data/string_ppi_v11_jacc_sim_mat.RDS')

# convert sim matrix to named list and remove all zeros
sim_list <- as.list(as.data.frame(sim))
names(sim_list) <- igraph::V(ppi)$name

condense <- function(sim_vec){
    names(sim_vec) <- igraph::V(ppi)$name
    ind <- which(sim_vec != 0)
    dense_vec <- sim_vec[ind]
}

sim_list_dense <- purrr::map(sim_list, ~condense(.))
saveRDS(sim_list_dense, 'data/string_ppi_v11_jacc_sim_list_dense.RDS')