add cohort colours and fix MZL

R/database.R

@@ -709,17 +709,17 @@ add_prps_result = function(incoming_metadata){
 #' icgc_metadata = add_icgc_metadata(incoming_metadata = my_meta)
 #'
 add_icgc_metadata = function(incoming_metadata){
-
-  #add trio metadata too!
-  trio_meta = "/projects/rmorin/projects/gambl-repos/gambl-rmorin/data/metadata/private_metadata/2021-04-30-DLBC_LSARP_Trios_with_metadata.tsv"
-  icgc_publ = suppressMessages(suppressWarnings(read_csv("/projects/rmorin/projects/gambl-repos/gambl-rmorin/data/metadata/raw_metadata/MALY_DE_tableS1.csv")))
+  repo_base = config::get("repo_base")
+  icgc_publ_file = paste0(repo_base,"data/metadata/raw_metadata/MALY_DE_tableS1.csv")
+  icgc_publ = suppressMessages(suppressWarnings(read_csv(icgc_publ_file)))
   icgc_publ = icgc_publ[,c(1:20)]
   #fix commas as decimals
   icgc_publ = mutate(icgc_publ, purity = str_replace(purity, ",", "."))
   icgc_publ = mutate(icgc_publ, sex = str_to_upper(sex))
-
-  icgc_raw = suppressMessages(read_tsv("/projects/rmorin/projects/gambl-repos/gambl-rmorin/data/metadata/raw_metadata/ICGC_MALY_seq_md.tsv"))
-
+
+  icgc_raw_path = paste0(repo_base,"data/metadata/raw_metadata/ICGC_MALY_seq_md.tsv")
+  icgc_raw = suppressMessages(read_tsv(icgc_raw_path))
+
   icgc_raw = icgc_raw %>%
     dplyr::select(-compression, -bam_available, -read_length, -time_point, -unix_group, -ffpe_or_frozen, -link_name)  %>%
     dplyr::filter(tissue_status %in% c("tumor", "tumour"))

R/utilities.R

@@ -834,7 +834,8 @@ collate_results = function(sample_table,
     output_file = config::get("table_flatfiles")$derived
     output_base = config::get("repo_base")
     output_file = paste0(output_base, output_file)
-    sample_table = read_tsv(output_file)
+    output_file = glue::glue(output_file)
+    sample_table = read_tsv(output_file) %>% dplyr::filter(sample_id %in% sample_table$sample_id)
   }else{
     message("Slow option: not using cached result. I suggest from_cache = TRUE whenever possible")
     #edit this function and add a new function to load any additional results into the main summary table
@@ -850,6 +851,7 @@ collate_results = function(sample_table,
   }
   if(write_to_file){
     output_file = config::get("table_flatfiles")$derived
+    output_file = glue(output_file)
     output_base = config::get("repo_base")
     output_file = paste0(output_base, output_file)
     write_tsv(sample_table, file = output_file)
@@ -1855,7 +1857,8 @@ get_gambl_colours = function(classification = "all",
       "SCBC"="#8c9c90",
       "UNSPECIFIED"="#cfba7c",
       "OTHER"="#cfba7c",
-      "MZL"="#065A7F"
+      "MZL"="#065A7F",
+      "SMZL"="#065A7F"
   )
   all_colours[["coo"]] = c(
     "ABC" = "#05ACEF",
@@ -1871,9 +1874,14 @@ get_gambl_colours = function(classification = "all",
     "DHITsigPos" = "#D62828",
     "NA" = "#ACADAF"
   )
-  all_colours[["cohort"]] = c("Chapuy"="#8B0000",
-                  "Arthur"= "#8845A8",
-                  "Schmitz"= "#2C72B2")
+  all_colours[["cohort"]] = c("Chapuy"="#8B0000","Chapuy, 2018"="#8B0000",
+                  "Arthur"= "#8845A8","Arthur, 2018"= "#8845A8",
+                  "Schmitz"= "#2C72B2","Schmitz, 2018"= "#2C72B2",
+                  "Reddy" = "#E561C3","Reddy, 2017" = "#E561C3",
+                  "Morin"= "#8DB753", "Morin, 2013"= "#8DB753",
+                  "Kridel"= "#4686B7", "Kridel, 2016"= "#4686B7",
+                  "ICGC"="#E09C3B","ICGC, 2018"="#E09C3B",
+                  "Grande"="#e90c8b", "Grande, 2019"="#e90c8b")
 
   all_colours[["indels"]] = c("DEL" = "#53B1FC", "INS" = "#FC9C6D")
 

config.yml

@@ -1,3 +1,7 @@
+remote:
+    project_base: "/Users/rmorin/gambl_results/"
+    repo_base: "/Users/rmorin/git/gambl/"
+
 default:
     project_base: "/projects/nhl_meta_analysis_scratch/gambl/results_local/"
     repo_base: "/projects/rmorin/projects/gambl-repos/gambl-rmorin/"
@@ -91,7 +95,7 @@ default:
         biopsies: "data/metadata/gambl_biopsy_metadata.tsv"
         samples: "data/metadata/gambl_samples_available.tsv"
         outcomes: "data/metadata/gambl_all_outcomes.tsv"
-        derived: "data/metadata/gambl_sample_results.tsv"
+        derived: "data/metadata/gambl_{seq_type_filter}_results.tsv"
 
     unmatched_normal_ids:
         gambl:
@@ -120,62 +124,3 @@ default:
                 hg38-nci: "04-24937N-Schmitz"
                 hg38: "BLGSP-71-06-00286-99A-01D"
                 hg19-clc: "PA011-G"
-
-testing:
-    project_base: "/projects/nhl_meta_analysis_scratch/gambl/results_local/"
-    repo_base: "/projects/rmorin/projects/gambl-repos/gambl-rmorin/"
-    derived_and_curated: "icgc_dart/derived_and_curated_metadata/"
-    database_name: "gambl_07_2021"
-    results_staging:
-        manta: "manta_current/level_3/01-gamblr_preprocess/"
-    results_merged:
-        tidy_expression_file: "/projects/rmorin/projects/gambl-repos/gambl-rmorin/results/icgc_dart/DESeq2-0.0_salmon-1.0/mrna--gambl-icgc-all/vst-matrix-Hugo_Symbol_tidy.tsv"
-        ex_matrix_file: "/projects/rmorin/projects/gambl-repos/gambl-rmorin/results/icgc_dart/DESeq2-0.0_salmon-1.0/mrna--gambl-icgc-all/vst-matrix-Hugo_Symbol.tsv.gz"
-
-#results from every combination of each of these will be included in the database
-    unix_groups: "gambl,icgc_dart"
-    genome_builds: "grch37,hg38,hs37d5"
-#the genome build to use for all coordinate-based results in the database
-    canonical_genome_build: "grch37"
-    analyses:
-        matched:
-            copy_number: "battenberg"
-            ssm: "slms-3"
-            sv: "manta"
-        unmatched:
-            copy_number: "controlfreec"
-            ssm: "slms-3"
-            sv: "manta"
-
-#tables that likely need to be populated sequentially or outside of R due to their size
-    results_tables:
-        ssm: "maf_slms3_hg19_icgc"
-        copy_number: "seg_battenberg_hg19"
-        sv: "bedpe_manta_hg19"
-        copy_number_unmatched: "seg_controlfreec_hg19"
-
-    results_filatfiles:
-        sv: "icgc_dart/cbioportal-1.0/01-filtered-manta/genome-grch37/all_merged_sv_info.tsv"
-        ssm:
-            all:
-                full: "icgc_dart/slms_3-1.0_vcf2maf-1.2/level_3/final_merged_grch37.maf"
-                cds: "icgc_dart/slms_3-1.0_vcf2maf-1.2/level_3/final_merged_grch37.CDS.maf"
-            gambl:
-                full: "gambl/slms_3-1.0_vcf2maf-1.2/level_3/final_merged_grch37.maf"
-                cds: "gambl/slms_3-1.0_vcf2maf-1.2/level_3/final_merged_grch37.CDS.maf"
-
-    results_directories:
-        manta: "manta_current/99-outputs/bedpe/"
-
-
-    tables:
-        biopsies: "biopsy_metadata"
-        samples: "sample_metadata"
-        outcomes: "outcome_metadata"
-        derived: "derived_data"
-
-    table_flatfiles:
-        biopsies: "data/metadata/gambl_biopsy_metadata.tsv"
-        samples: "data/metadata/gambl_samples_available.tsv"
-        outcomes: "data/metadata/gambl_all_outcomes.tsv"
-        derived: "data/metadata/gambl_sample_results.tsv"

test_remote.R

@@ -0,0 +1,15 @@
+
+setwd("~/git/gambl/") # set this path to point to your local clone of the gambl repo
+# in your config.yml file set repo_base under the remote config (at the top) to point to this path also
+library(GAMBLR)
+Sys.setenv(R_CONFIG_ACTIVE= "remote")
+
+
+config::get("project_base")
+#[1] "/Users/rmorin/gambl_results/"
+
+all_meta = get_gambl_metadata()
+
+collated = collate_results(from_cache=TRUE)
+
+collated = collate_results(from_cache=TRUE,seq_type_filter="capture")