Start implementing ciri_de

src/ciri_de.R

@@ -0,0 +1,37 @@
+
+
+work_dir <- "../data/work"
+circ_dir <- "../data/circ"
+stringtie_dir <- "../data/transcripts"
+
+create_df <- function(samples, condition) {
+  ciri_files <- file.path(circ_dir, paste0(samples, ".gtf"))
+  stringtie_files <- file.path(stringtie_dir, paste0(samples, "_out.gtf"))
+  conditions <- rep(condition, length(samples))
+  return(data.frame("ciri" = ciri_files, "condition" = conditions, "stringtie" = stringtie_files, row.names = samples))
+}
+
+ciri_prepde <- function(df, wd) {
+  samplesheet <- file.path(wd, "samplesheet_ciri.tsv")
+  write.table(df, samplesheet, sep = "\t", quote = FALSE, row.names = TRUE, col.names = FALSE)
+
+  lib_file <- file.path(wd, "lib_ciri.csv")
+  circ_file <- file.path(wd, "circ_ciri.csv")
+  bsj_file <- file.path(wd, "bsj_ciri.csv")
+  ratio_file <- file.path(wd, "ratio_ciri.csv")
+
+  system(paste("prep_CIRIquant -i", samplesheet, "--lib", lib_file, "--circ", circ_file, "--bsj", bsj_file, "--ratio", ratio_file))
+
+  return(list(lib_file, circ_file, bsj_file, ratio_file))
+}
+
+run <- function(control, treatment) {
+  control_df <- create_df(control, "C")
+  treatment_df <- create_df(treatment, "T")
+  df <- rbind(control_df, treatment_df)
+
+  temp_dir <- work_dir
+
+  p_ciri <- ciri_prepde(df[c("ciri", "condition")], temp_dir)
+
+}

src/statistics.R

@@ -77,5 +77,19 @@ statisticsServer <- function(id, expr_matrix, phenotype) {
         choices = possible_vals()
       )
     })
+
+    test_result <- eventReactive(input$run_test, {
+      print("Running test")
+      column <- input$col
+      control <- input$control
+      treatment <- input$treatment
+
+      cur_phenotype <- phenotype()
+
+      control_ids <- which(cur_phenotype[[column]] %in% control)
+      treatment_ids <- which(cur_phenotype[[column]] %in% treatment)
+
+
+    })
   })
 }

test.R

@@ -1,24 +1,11 @@
 source("load.R")
-library(umap)
+source("ciri_de.R")
 
 
-circ <- loadCirc()
 phenotype <- loadPhenotype()
 
-circ <- log1p(circ)
+control_samples <- rownames(head(phenotype))
+treatment_samples <- rownames(tail(phenotype))
 
-pca10 <- prcomp(t(circ), rank. = 10)
+run(control_samples, treatment_samples)
 
-data <- pca10$x
-
-res <- umap(data,
-  n_components = 3,
-  n_neighbors = min(15, nrow(data)) - 1
-)
-
-layout <- res[["layout"]]
-layout <- data.frame(layout)
-
-print(dim(layout))
-
-cbind(layout, phenotype)