adding hisens purity to facets-qc output. also compile_cohort_annotat…

…ions now includes manual review info
mskcc · Jun 6, 2020 · 5f76a73 · 5f76a73
1 parent 65236ec
commit 5f76a73
Show file tree

Hide file tree

Showing 5 changed files with 39 additions and 5 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,6 +1,6 @@
 Package: facetsPreview
 Title: Review and refit FACETS runs
-Version: 2.1.2
+Version: 2.1.4
 Authors@R: person("First", "Last", email = "[email protected]", role = c("aut", "cre"))
 Description: Review and refit FACETS runs
 biocViews: 

diff --git a/R/compile_cohort_annotations.R b/R/compile_cohort_annotations.R
@@ -42,8 +42,9 @@ compile_cohort_annotations <- function(samples_to_annotate, output_prefix, ncore
 
             ## read from facets_qc.txt
             qc_file = paste0(sample_path, '/facets_qc.txt')
+            review_file = paste0(sample_path, '/facets_review.manifest')
 
-            if (!file.exists(qc_file)) { return() }
+            if (!file.exists(qc_file)| !file.exists(review_file)) { return() }
 
             qc_runs = fread(qc_file) %>% filter(fit_name != 'Not selected')
 
@@ -58,6 +59,32 @@ compile_cohort_annotations <- function(samples_to_annotate, output_prefix, ncore
               fit = qc_runs %>% head(n=1)
             }
 
+            reviews <-
+              fread(review_file) %>%
+              filter(fit_name == fit$fit_name) %>%
+              arrange(desc(date_reviewed)) %>%
+              select(fit_name, review_notes, reviewed_by, date_reviewed, use_only_purity_run, use_edited_cncf, reviewer_set_purity) %>%
+              head(n=1)
+
+            fit <-
+              fit %>% 
+              left_join(reviews) 
+
+            ####
+            #### update purity and ploidy based on reviews;
+            ####
+            fit <- 
+              fit %>% 
+              rowwise %>%
+              mutate(purity = ifelse(!is.na(reviewer_set_purity) & reviewer_set_purity != '', 
+                                     reviewer_set_purity,
+                                     ifelse(!is.na(use_only_purity_run) & use_only_purity_run, 
+                                            purity_run_Purity, 
+                                            hisens_run_Purity))) %>%
+              mutate(ploidy = ifelse(!is.na(use_only_purity_run) & use_only_purity_run, 
+                                     purity_run_Ploidy, 
+                                     hisens_run_Ploidy))
+
             return (fit)
           }, .parallel = parallelize)
     )

diff --git a/R/generate_genomic_annotations.R b/R/generate_genomic_annotations.R
@@ -106,8 +106,14 @@ generate_genomic_annotations = function(sample_id, sample_path, config_file, reg
     write.table(qc, paste0(prefix, '.qc.txt'), quote=F, row.names=F, sep='\t')
 
     # Write gene level // use hisensitivity run
-    gene_level = facetsSuite::gene_level_changes(hisens_output, 'hg19') %>%
-      add_column(sample = sample_id, .before = 1)
+
+    if (r$use_only_purity_run) {
+      gene_level = facetsSuite::gene_level_changes(purity_output, 'hg19', 'em')
+    } else {
+      gene_level = facetsSuite::gene_level_changes(hisens_output, 'hg19', 'em')
+    }
+    gene_level = gene_level %>% add_column(sample = sample_id, .before = 1)
+
     write.table(gene_level, paste0(prefix, '.gene_level.txt'), quote=F, row.names=F, sep='\t')
 
     # Write arm level // use purity run

diff --git a/R/global.R b/R/global.R
@@ -231,6 +231,7 @@ metadata_init <- function(sample_id, sample_path, progress = NULL, update_qc_fil
                                       hisens_run_snp_nbhd = get0("hisens_snp.nbhd", ifnotfound = NA),
                                       hisens_run_ndepth = get0("hisens_ndepth", ifnotfound = NA),
                                       hisens_run_hisens = round_down(get0("hisens_hisens", ifnotfound = NA)),
+                                      hisens_run_Purity = round_down(get0("hisens_Purity", ifnotfound = NA)),
                                       hisens_run_Ploidy = round_down(get0("hisens_Ploidy", ifnotfound = NA)),
                                       hisens_run_dipLogR = round_down(get0("hisens_dipLogR", ifnotfound = NA)),
 

diff --git a/example_facets_preview.json b/example_facets_preview.json
@@ -30,7 +30,7 @@
   ],
   "verify_sshfs_mount" : "juno",
   "r_script_path" : "/opt/common/CentOS_7-dev/bin/Rscript",
-  "facets_suite_lib": "/juno/work/ccs/shared/software/R_libs/facetsSuite/2.0.5/",
+  "facets_suite_lib": "/juno/work/ccs/shared/software/R_libs/facetsSuite/2.0.6/",
   "facets_suite_run_wrapper": "/juno/work/ccs/shared/software/R_libs/facetsSuite/2.0.4/run-facets-wrapper.R",
   "facets_qc_script": "/juno/work/ccs/bandlamc/git/facets-preview/facets_qc/v1.0/facets_fit_qc.R"
 }