Biomarker matching fixes

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "2.3.1.9001"
+current_version = "2.3.1.9002"
 search = "{current_version}"
 replace = "{new_version}"
 message = "Bump version: {current_version} → {new_version}"

.github/workflows/build_conda_recipes.yaml

@@ -12,7 +12,7 @@ env:
   conda_env_yaml: conda/env/yaml
   conda_env_lock: conda/env/lock
   conda_org: pcgr
-  VERSION: '2.3.1.9001' # bump
+  VERSION: '2.3.1.9002' # bump
 jobs:
   conda_build:
     # When merging to one of the branches above and the commit message matches

conda/env/yaml/pcgr.yaml

@@ -6,7 +6,7 @@ channels:
   - bioconda
 
 dependencies:
-  - pcgr ==2.3.1.9001 # bump
+  - pcgr ==2.3.1.9002 # bump
   - bioconda::bedtools
   - bioconda::bcftools
   - bioconda::ensembl-vep ==115.1

conda/env/yaml/pcgrr.yaml

@@ -7,7 +7,7 @@ channels:
 
 dependencies:
   - python
-  - r-pcgrr ==2.3.1.9001 # bump
+  - r-pcgrr ==2.3.1.9002 # bump
   - r-cpsr
   - r-argparse
   - bioconductor-bsgenome.hsapiens.ucsc.hg38

conda/env/yaml/pkgdown.yaml

@@ -4,7 +4,7 @@ channels:
   - bioconda
   - conda-forge
 dependencies:
-  - r-pcgrr ==2.3.1.9001 # bump
+  - r-pcgrr ==2.3.1.9002 # bump
   - r-pkgdown
   - r-readr
   - r-glue

conda/recipe/pcgr/recipe.yaml

@@ -1,6 +1,6 @@
 context:
   name: pcgr
-  version: 2.3.1.9001 # bump
+  version: 2.3.1.9002 # bump
 
 package:
   name: ${{ name|lower }}

conda/recipe/pcgrr/recipe.yaml

@@ -1,6 +1,6 @@
 context:
   name: r-pcgrr
-  version: 2.3.1.9001 # bump
+  version: 2.3.1.9002 # bump
 
 package:
   name: ${{ name|lower }}

pcgr/_version.py

@@ -1 +1 @@
-__version__ = '2.3.1.9001' # bump
+__version__ = '2.3.1.9002' # bump

pcgrr/DESCRIPTION

@@ -1,7 +1,7 @@
 Package: pcgrr
 Type: Package
 Title: Personal Cancer Genome ReporteR
-Version: 2.3.1.9001
+Version: 2.3.1.9002
 Authors@R:
     c(person(given = "Sigve",
              family = "Nakken",

pcgrr/NAMESPACE

@@ -164,6 +164,7 @@ export(stats_report_germline)
 export(stats_report_snv_indel)
 export(stats_type_snv_indel)
 export(strip_html)
+export(sync_biomarker_evidence)
 export(tcga_somatic_status)
 export(tier_af_distribution)
 export(tmb_doc_note)

pcgrr/R/input_data.R

@@ -372,6 +372,19 @@ load_somatic_snv_indel <- function(
     }
   }
 
+  ## Re-sync biomarker evidence items/classifications with the variant set
+  ## as it stands after all filtering above (allelic depth/fraction, and -
+  ## for tumor-only input - germline/non-exonic filtering). Biomarker
+  ## matching was performed once, early, inside load_dna_variants(), against
+  ## the variant set as it existed prior to this filtering - without this
+  ## step, evidence items for variants removed above remain orphaned in
+  ## 'bm_evidence' (see https://github.com/sigven/pcgr/issues/302)
+  if ("bm_evidence" %in% names(callset)) {
+    callset[['bm_evidence']] <- sync_biomarker_evidence(
+      bm_evidence = callset[['bm_evidence']],
+      var_df = callset[['variant']])
+  }
+
   if (NROW(callset[['variant']]) > 0) {
     callset[['variant']] <- callset[['variant']] |>
       dplyr::arrange(

pcgrr/R/variant_classification.R

@@ -281,6 +281,82 @@ assign_amp_asco_cap_tiers <- function(
 
 }
 
+#' Re-synchronize biomarker evidence items/classifications with a
+#' (possibly further-filtered) variant set
+#'
+#' Biomarker evidence matching (\code{map_biomarker_data()},
+#' \code{assign_amp_asco_cap_tiers()}) is performed once, early, against
+#' the variant set as it existed right after annotation. Callers may go on
+#' to remove variants from that set afterwards (e.g. allelic depth/fraction
+#' filtering via \code{filter_read_support()}, or germline/non-exonic
+#' filtering for tumor-only input). Without re-syncing, evidence items for
+#' variants removed by such downstream filtering remain "orphaned" -
+#' present in \code{bm_evidence} but absent from the variant set - which
+#' surfaces as biomarker-matched variants missing from the variant listing
+#' in the report while still appearing in the biomarker evidence listing
+#' (see <https://github.com/sigven/pcgr/issues/302>).
+#'
+#' @param bm_evidence list with biomarker evidence data (as initialized by
+#' \code{init_biomarker_content()}), i.e. top-level 'eitems'/'classification'
+#' data frames plus one sub-list per clinical significance category, each
+#' with its own 'eitems'/'classification' data frames
+#' @param var_df data frame with the final (filtered) variant set
+#'
+#' @return bm_evidence list, with all 'eitems'/'classification' data frames
+#' limited to records matching a variant in var_df
+#'
+#' @export
+sync_biomarker_evidence <- function(bm_evidence = NULL, var_df = NULL) {
+
+  invisible(assertthat::assert_that(
+    is.list(bm_evidence),
+    msg = "Argument 'bm_evidence' needs to be of type list"))
+  invisible(assertthat::assert_that(
+    is.data.frame(var_df),
+    msg = "Argument 'var_df' needs to be of type data.frame"))
+
+  join_cols <- intersect(
+    c("VAR_ID", "VARIANT_CLASS", "ENTREZGENE"),
+    colnames(var_df))
+
+  if (length(join_cols) == 0 || NROW(var_df) == 0) {
+    return(bm_evidence)
+  }
+
+  variant_keys <- var_df |>
+    dplyr::select(dplyr::all_of(join_cols)) |>
+    dplyr::distinct()
+
+  sync_df <- function(df) {
+    if (is.data.frame(df) && NROW(df) > 0 &&
+        all(join_cols %in% colnames(df))) {
+      return(dplyr::semi_join(df, variant_keys, by = join_cols))
+    }
+    return(df)
+  }
+
+  for (elem in c("eitems", "classification")) {
+    if (elem %in% names(bm_evidence)) {
+      bm_evidence[[elem]] <- sync_df(bm_evidence[[elem]])
+    }
+  }
+
+  for (clnsig in names(bm_evidence)) {
+    if (is.list(bm_evidence[[clnsig]]) &&
+        !is.data.frame(bm_evidence[[clnsig]])) {
+      for (elem in c("eitems", "classification")) {
+        if (elem %in% names(bm_evidence[[clnsig]])) {
+          bm_evidence[[clnsig]][[elem]] <-
+            sync_df(bm_evidence[[clnsig]][[elem]])
+        }
+      }
+    }
+  }
+
+  return(bm_evidence)
+
+}
+
 #' Assign tiers of clinical significance (AMP/ASCO/CAP framework) to
 #' somatic CNAs
 #'

pcgrr/vignettes/installation.Rmd

@@ -15,7 +15,7 @@ require(glue, include.only = "glue")
 
 ```{r vars, echo=FALSE}
 Sys.setenv(VEP_VERSION = "115")
-Sys.setenv(PCGR_VERSION = "2.3.1.9001")
+Sys.setenv(PCGR_VERSION = "2.3.1.9002")
 Sys.setenv(BUNDLE_VERSION = "20260620")
 VEP_VERSION <- Sys.getenv("VEP_VERSION")
 PCGR_VERSION <- Sys.getenv("PCGR_VERSION")

pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "pcgr"
-version = "2.3.1.9001" # bump
+version = "2.3.1.9002" # bump
 description = "Personal Cancer Genome Reporter (PCGR) - variant interpretation for precision cancer medicine"
 authors = [
     {name = "Sigve Nakken", email = "sigven@gmail.com"},