Reduce memory footprint during analysis

scripts/analyze.R

@@ -7,7 +7,7 @@ library(here)
 
 i_am("scripts/input.R")
 
-data <- fread(here("scripts", "input", "data_long.csv.gz"))
+data <- fread(here("scripts", "input", "data_long.csv"))
 
 data[, `:=`(
     expression_median = median(expression),

scripts/input.R

@@ -70,21 +70,24 @@ getpm <- DGEList(counts = read_counts) |>
 data_wide_samples <- data.table(getpm, keep.rownames = "gene")
 data_wide_samples[, hgnc_symbol := hgnc_symbols]
 
+# Create lookup tables for genes and samples.
+
+genes <- data_wide_samples[, .(id = .I, gene, hgnc_symbol)]
+fwrite(genes, file = here("scripts", "input", "genes.csv"))
+
+sample_names <- colnames(data_wide_samples[, !c("gene", "hgnc_symbol")])
+samples <- data.table(id = seq_along(sample_names), sample = sample_names)
+fwrite(samples, file = here("scripts", "input", "samples.csv"))
+
+data_wide_samples[, `:=`(gene = .I, hgnc_symbol = NULL)]
+colnames(data_wide_samples) <- c("gene", seq_along(sample_names))
+
 data_long <- melt(
   data_wide_samples,
-  id.vars = c("gene", "hgnc_symbol"),
+  id.vars = "gene",
   variable.name = "sample",
   value.name = "expression",
   variable.factor = FALSE
 )
 
-fwrite(
-  data_wide_samples,
-  file = here(
-    "scripts",
-    "input",
-    "data_wide_samples.csv.gz"
-  )
-)
-
-fwrite(data_long, file = here("scripts", "input", "data_long.csv.gz"))
+fwrite(data_long, file = here("scripts", "input", "data_long.csv"))

scripts/ranking.R

@@ -6,6 +6,7 @@ library(here)
 
 i_am("scripts/input.R")
 
+genes <- fread(here("scripts", "input", "genes.csv"))
 data <- fread(here("scripts", "output", "results.csv"))
 
 data[, score := 0.5 * above_95 +
@@ -22,17 +23,24 @@ data[is.na(score), score := 0.0]
 
 setorder(data, -score)
 
+# Reintroduce gene IDs and HGNC symbols.
+
+setnames(data, "gene", "id")
+
+data <- merge(
+  data,
+  genes,
+  by = "id",
+  all.x = TRUE,
+  sort = FALSE
+)
+
+setnames(data, "hgnc_symbol", "hgnc_name")
+data[, id := NULL]
+
 # Remove duplicates. This will keep the best row for each duplicated gene.
 data <- unique(data, by = "gene")
 
-data[, `:=`(
-  hgnc_name = gprofiler2::gconvert(
-    gene,
-    target = "HGNC",
-    mthreshold = 1,
-    filter_na = FALSE
-  )$target,
-  rank = .I
-)]
+data[, rank := .I]
 
 fwrite(data, file = here("scripts", "output", "genes.csv"))