Merge pull request #135 from umccr/prod-vs-dev-comparisons

Prod vs dev comparisons

inst/scripts/compare.R

@@ -0,0 +1,74 @@
+require(dplyr)
+require(readr)
+require(dracarys)
+require(here)
+require(glue)
+require(tibble)
+require(tidyr)
+
+# gds results per subject
+o <- here("nogit/comparisons")
+tokpro <- Sys.getenv("ICA_ACCESS_TOKEN_PRO")
+tokdev <- Sys.getenv("ICA_ACCESS_TOKEN_DEV")
+s <- tibble::tribble(
+  ~sbj, ~pro, ~dev,
+  "SBJ04468", "20231212b3b22850/SBJ04468__L2301504/PRJ231328.results", "output/SBJ04468__L2301504/PRJ231328.results",
+  "SBJ04469", "20231209eeca1fe7/SBJ04469__L2301505/MDX230556.results", "output/SBJ04469__L2301505/MDX230556.results",
+  "SBJ04578", "2024011447748eb1/SBJ04578__L2400043/MDX230546.results", "output_test/SBJ04578__L2400043/MDX230546.results"
+) |>
+  mutate(
+    pro = glue("gds://production/analysis_data/{sbj}/rnasum/{pro}"),
+    dev = glue("gds://development/sehrish/rnasum_0.5.0/{sbj}/{dev}")
+  ) |>
+  pivot_longer(cols = c("pro", "dev"), names_to = "namespace", values_to = "gdsdir") |>
+  mutate(
+    outdir = file.path(o, sbj, namespace),
+    token = ifelse(namespace == "pro", tokpro, tokdev)
+  )
+
+# download results
+regex <- tibble::tribble(
+  ~regex, ~fun,
+  "^genes\\.expr\\.perc\\.html$", "foo1",
+  "^genes\\.expr\\.z\\.html$", "foo2",
+)
+# try one first
+# dracarys::dr_gds_download(gdsdir = s$gdsdir[1], outdir = s$outdir[1], token = s$token[1], page_size = 200, dryrun = TRUE, regexes = regex)
+x <- s |>
+  rowwise() |>
+  mutate(
+    dl = list(dracarys::dr_gds_download(gdsdir = gdsdir, outdir = outdir, token = token, page_size = 200, dryrun = FALSE, regexes = regex))
+  ) |>
+  ungroup()
+
+# after download, open html and export csv
+d <- s |>
+  mutate(
+    exprp = file.path(outdir, "perc.csv"),
+    exprz = file.path(outdir, "z.csv")
+  ) |>
+  select(sbj, namespace, exprp, exprz) |>
+  pivot_longer(cols = c("exprp", "exprz"), names_to = "ftype", values_to = "fpath") |>
+  pivot_wider(names_from = "namespace", values_from = "fpath")
+
+
+i <- 6
+dev <- readr::read_csv(d$dev[i])
+pro <- readr::read_csv(d$pro[i])
+
+# now explore expression differences in reference and patient columns
+# between dev and prod.
+dplyr::left_join(dev, pro, by = "Gene", suffix = c(".dev", ".pro")) |>
+  dplyr::mutate(
+    # Ref_equal = `KIRP (TCGA).dev` == `KIRP (TCGA).pro`,
+    # Ref_equal = `PANCAN (TCGA).dev` == `PANCAN (TCGA).pro`,
+    # Ref_equal = `PAAD (TCGA).dev` == `PAAD (TCGA).pro`,
+    Pat_equal = Patient.dev == Patient.pro,
+    # Ref_diff = abs(`PANCAN (TCGA).dev` - `PANCAN (TCGA).pro`),
+    # Ref_diff = abs(`PAAD (TCGA).dev` - `PAAD (TCGA).pro`),
+    Pat_diff = abs(Patient.dev - Patient.pro)
+  ) |>
+  dplyr::select(Gene, contains("KIRP"), contains("PANCAN"), contains("Patient"), contains("diff"), everything()) |>
+  dplyr::filter(Pat_diff > 0) |>
+  dplyr::arrange(desc(Pat_diff)) |>
+  dplyr::arrange(desc(Ref_diff))