Handle full join fail during bambu quant [CW-7333]

CHANGELOG.md

@@ -5,6 +5,15 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 
+## [v2.0.1]
+
+This patch release of `wf-transcriptomes` handles an additional quantification failure case that was not observed before release.
+Users of wf-transcriptomes v2.0.0 who have encountered issues during quantification should adopt this release.
+
+### Fixed
+- Error in full_join encountered during runPerSampleBambuQuant when all read classes have no compatible transcript assignment.
+
+
 ## [v2.0.0]
 
 This release refreshes `wf-transcriptomes` around a new reference-guided transcriptomics workflow built on `bambu`, with `SQANTI3` transcript classification and QC, `DESeq2` for differential gene expression, `DEXSeq` for differential transcript usage, and per-sample modified base summarisation with `modkit` when modification tags are present in aligned BAMs.

bin/workflow_glue_r/R/bambu.R

@@ -325,6 +325,9 @@ bambu_known_quant_edge_error_kind <- function(msg) {
     if (grepl("eqClassById` with `y$eqClassById` due to incompatible types.", msg, fixed = TRUE)) {
         return("eqClassById_incompatible_types")
     }
+    if (grepl("Can't convert `x$txid` <vctrs_unspecified> to match type of `txid` <integer>.", msg, fixed = TRUE)) {
+        return("txid_unspecified_incompatible_types")
+    }
     NA_character_
 }
 

bin/workflow_glue_r/tests/testthat/test_bambu.R

@@ -663,6 +663,62 @@ testthat::test_that("quant mode catches known eqClassById incompatible-type edge
     testthat::expect_equal(colnames(se), c("sampleA"))
 })
 
+testthat::test_that("quant mode catches known txid unspecified edge case and writes empty outputs", {
+    fixture_dir <- tempfile("bambu-quant-known-txid-edge-")
+    dir.create(fixture_dir)
+
+    chunk_bundle <- list(
+        chunk_id = "chr2",
+        seqname = "chr2",
+        aliases = c("sampleA"),
+        sample_df = data.frame(alias = c("sampleA"), stringsAsFactors = FALSE),
+        rc_files = list(sampleA = make_test_tx_se(sample_names = "sampleA")),
+        annotation_tx_count = 1L
+    )
+    chunk_rds <- file.path(fixture_dir, "chr2.rds")
+    saveRDS(chunk_bundle, chunk_rds)
+
+    discovered_annotation_rds <- file.path(fixture_dir, "annotations.rds")
+    saveRDS(make_test_bambu_row_ranges(fixture_dir), discovered_annotation_rds)
+
+    analysis_called <- FALSE
+    fake_analysis <- function(...) {
+        analysis_called <<- TRUE
+        stop(
+            paste0(
+                "Can't convert `x$txid` <vctrs_unspecified> ",
+                "to match type of `txid` <integer>."
+            ),
+            call. = FALSE
+        )
+    }
+
+    args <- workflow_glue_r_normalise_args(
+        list(
+            mode = "quant",
+            genome = "genome.fa",
+            out_dir = file.path(fixture_dir, "out"),
+            chunk_rds = chunk_rds,
+            discovered_annotation_rds = discovered_annotation_rds,
+            transcriptome_mode = "discover",
+            ndr = NULL,
+            threads = 2
+        ),
+        bambu_arg_spec()
+    )
+
+    testthat::expect_warning(
+        suppressMessages(main_run_bambu(args, analysis_fn = fake_analysis)),
+        "known bambu chunk edge case"
+    )
+
+    testthat::expect_true(analysis_called)
+
+    se <- readRDS(file.path(args$out_dir, "bambu_transcripts.rds"))
+    testthat::expect_equal(nrow(se), 0)
+    testthat::expect_equal(colnames(se), c("sampleA"))
+})
+
 testthat::test_that("empty mode writes valid empty outputs including bambu rds files", {
     fixture_dir <- tempfile("bambu-empty-mode-")
     dir.create(fixture_dir)

nextflow.config

@@ -67,7 +67,7 @@ manifest {
     description     = 'Long-read transcript discovery, quantification, differential expression, QC and mod counting.'
     mainScript      = 'main.nf'
     nextflowVersion = '>=23.04.2'
-    version         = 'v2.0.0'
+    version         = 'v2.0.1'
 }
 
 process {