cRegulome.db

1 Libraries

library(cRegulome)
library(RSQLite)
library(ggplot2)
library(dplyr)
library(R.utils)

2 Database

file <- "cRegulome.db"

if(!file.exists(file)){
  downloader::download("https://s3-eu-west-1.amazonaws.com/pfigshare-u-files/9537385/cRegulome.db.gz","cRegulome.db.gz")
  gunzip(paste0(file,".gz"), remove = FALSE)
}
# connect to the db file
conn <- dbConnect(SQLite(), file)

3 Checking database

# check tables
dbListTables(conn)

## [1] "cor_mir"     "cor_tf"      "targets_mir" "targets_tf"

# check fields
dbListFields(conn,"targets_tf")

## [1] "tf"      "feature" "study"

# available studies
dbListFields(conn, "cor_tf")

##  [1] "tf"         "feature"    "ACC"        "BLCA"       "BRCA_1"    
##  [6] "BRCA_2"     "CESC"       "COAD_READ*" "GBM*"       "HNSC"      
## [11] "KICH"       "KIRC"       "KIRP"       "LAML"       "LGG"       
## [16] "LIHC"       "LUAD"       "LUSC"       "MESO"       "OV"        
## [21] "PAAD"       "PCPG"       "PRAD"       "SARC"       "SKCM"      
## [26] "STES*"      "TGCT"       "THCA"       "THYM"       "UCEC"      
## [31] "UVM"

4 Finding targets for ACC

TF_list <- TCGAbiolinks:::EAGenes %>% 
  filter(Family %in% "transcription regulator") %>% 
  pull(Gene) %>% 
  as.character()


# Manual function
creg_tf <- get_tf(conn,
                  tf = TF_list[1:300],
                  study = "ACC",
                  targets_only = TRUE)
creg_tf

# Manual function
tf <- TF_list[1:300]
study.tf <- tbl(conn, "targets_tf") %>% 
  filter(study == 'ACC') %>% 
  pull(tf) %>% 
  unique
tf <- tf[tf %in% study.tf]
study <- "ACC"

ll1 <- list()
ll2 <- list()
for (m in 1:length(tf)) {
  for (s in 1:length(study)) {
    tars_stat <- stat_make_targets(reg = tf[m], study = study[s], type = "tf")
    tars <- stat_collect_targets(conn, stat = tars_stat)
    stat <- stat_make(tf[m], study = study[s], targets = tars, type = "tf")
    df <- stat_collect(conn, study = study[s], stat,type = "tf")
    ll2[[s]] <- df
  }
  ll1[[m]] <- ll2
}
ll <- unlist(ll1, recursive = FALSE)
dat <- do.call("rbind", ll)
dat$cor <- dat$cor/100
dat <- na.omit(dat)

5 Finding targets for COAD-READ

TF_list <- TCGAbiolinks:::EAGenes %>% 
  filter(Family %in% "transcription regulator") %>% 
  pull(Gene) %>% 
  as.character()

# get available TF for study
study.tf <- tbl(conn, "targets_tf") %>% 
  filter(study == 'COAD_READ*') %>% 
  pull(tf) %>% 
  unique

# find the TF within database table
idx <- which(TF_list %in% study.tf)

results <- plyr::adply(TF_list[idx][1:3],
                       .margins = 1,
                       .fun = function(x){
                         # get tf targets
                         tars <- tbl(conn, "targets_tf")  %>% 
                           filter(tf == x & study ==  "COAD_READ*") %>% 
                           pull(2)
                         
                         tbl(conn, "cor_tf") %>% 
                           select(c("tf", "feature", "COAD_READ*")) %>% 
                           filter(tf == x & feature %in% tars) %>% 
                           as.data.frame
                       },.id = NULL)
results

results <- results %>% reshape2::melt()

## Using tf, feature as id variables

colnames(results) <- c("tf", "feature", "study","cor")
results$cor <- results$cor/100
results <- na.omit(results)
results