kfukuta/wf-transcriptomes-v202
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'gene_names' into 'dev'

Browse Source 
Add gene name to TSV outputs [CW-7160]

See merge request epi2melabs/workflows/wf-transcriptomes!274
This commit is contained in:
Sam Nicholls 2026-05-22 10:23:21 +00:00
commit 89626ab016
7 changed files with 278 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
bin/workflow_glue_r/R/bambu.R
View File

@ -565,6 +565,7 @@ bambu_run_collate_mode <- function(args, gene_expression_fn, write_gtf_fn) {
out_dir = args$out_dir,
transcriptome_mode = args$transcriptome_mode,
ndr = args$ndr,
annotation = args$annotation,
gene_expression_fn = gene_expression_fn,
write_gtf_fn = write_gtf_fn
)
@ -1038,6 +1039,7 @@ bambu_collate_chunk_outputs <- function(
out_dir,
transcriptome_mode = "discover",
ndr = NULL,
annotation = NULL,
gene_expression_fn = bambu::transcriptToGeneExpression,
write_gtf_fn = bambu::writeToGTF
) {
@ -1077,7 +1079,8 @@ bambu_collate_chunk_outputs <- function(
args <- list(
out_dir = out_dir,
transcriptome_mode = transcriptome_mode,
ndr = ndr
ndr = ndr,
annotation = annotation
)
bambu_write_outputs(
se,
@ -1111,6 +1114,62 @@ bambu_write_matrix_tsv <- function(se_obj, assay_name, id_col, meta_df, output_p
invisible(gc(verbose = FALSE))
}
#' Enrich SummarizedExperiment rowData with annotation-derived names.
#'
#' Reads `gene_name` and `transcript_name` from a GFF annotation file
#' and adds them as columns in the `rowData` of the transcript-level and
#' gene-level SummarizedExperiment objects. Matching is performed on
#' `GENEID`or TXNAME
#' If `annotation` is missing/NULL, or if the GTF does not contain the
#' relevant attributes, the objects are returned unchanged.
#'
#' @param se A `SummarizedExperiment` of transcript-level counts
#' Must have GENEID in rowData for gene name enrichment.
#' @param gene_se A SummarizedExperiment of gene-level counts (bambu
#' Must have GENEID in rowData or use rownames as gene IDs.
#' @param annotation Path to a GTF/GFF annotation file, or NULL. When
#' NULL or the file is absent the function is a no-op.
#'
#' @return A named list with elements `se`` and `gene_se`, each being
#' the (possibly enriched) input object.
bambu_add_annotation_names <- function(se, gene_se, annotation) {
feature_maps <- workflow_glue_r_annotation_name_maps(annotation)
gene_name_map <- feature_maps$gene
transcript_name_map <- feature_maps$transcript
if (nrow(gene_name_map) < 1 && nrow(transcript_name_map) < 1) {
return(list(se = se, gene_se = gene_se))
}
tx_row_data <- SummarizedExperiment::rowData(se)
update_tx_row_data <- FALSE
if (nrow(gene_name_map) > 0 && "GENEID" %in% names(tx_row_data)) {
tx_row_data$gene_name <- gene_name_map$gene_name[
match(as.character(tx_row_data$GENEID), gene_name_map$GENEID)
]
update_tx_row_data <- TRUE
}
if (nrow(transcript_name_map) > 0 && "TXNAME" %in% names(tx_row_data)) {
tx_row_data$transcript_name <- transcript_name_map$transcript_name[
match(as.character(tx_row_data$TXNAME), transcript_name_map$TXNAME)
]
update_tx_row_data <- TRUE
}
if (update_tx_row_data) {
SummarizedExperiment::rowData(se) <- S4Vectors::DataFrame(tx_row_data)
}
gene_row_data <- SummarizedExperiment::rowData(gene_se)
if (nrow(gene_name_map) > 0 && "GENEID" %in% names(gene_row_data)) {
gene_row_data$gene_name <- gene_name_map$gene_name[
match(as.character(gene_row_data$GENEID), gene_name_map$GENEID)
]
SummarizedExperiment::rowData(gene_se) <- S4Vectors::DataFrame(gene_row_data)
}
list(se = se, gene_se = gene_se)
}
bambu_write_outputs <- function(
se,
gene_se,
@ -1120,6 +1179,10 @@ bambu_write_outputs <- function(
write_gtf_fn = bambu::writeToGTF,
write_rds = TRUE
) {
outputs <- bambu_add_annotation_names(se, gene_se, args$annotation)
se <- outputs$se
gene_se <- outputs$gene_se
row_ranges <- SummarizedExperiment::rowRanges(se)
if (length(row_ranges) > 0) {
write_gtf_fn(

76
bin/workflow_glue_r/R/common.R
View File

@ -53,3 +53,79 @@ workflow_glue_r_empty_tsv <- function(columns) {
names(out) <- columns
out
}
#' Extract a deduplicated id-to-name mapping from a GFF annotation file.
#'
#' Uses annotation metadata object from rtracklayer to generate a two-column
#' `data.frame` mapping feature IDs to display names.
#' When a feature ID maps to multiple names only the first observed name is kept.
#'
#' @param annotation_path data.frame with annotation metadata
#' @param id_column Name of the GFF attribute to use as the identifier
#' (e.g. `gene_id` or `transcript_id`).
#' @param name_column Name of the GFF attribute to use as the display name
#' (e.g. `gene_name` or `transcript_name`).
#' @param output_id_column Column name for the identifier in the returned
#' `data.frame` (e.g. `GENEID` or `TXNAME`).
#'
#' @return A `data.frame` with columns `output_id_column` and `name_column`.
#' Returns an empty `data.frame` with those columns if the required
#' attributes are absent or all values are missing.
workflow_glue_r_annotation_name_map_from_meta <- function(
annotation_meta,
id_column,
name_column,
output_id_column
) {
output_columns <- c(output_id_column, name_column)
if (!all(c(id_column, name_column) %in% names(annotation_meta))) {
return(workflow_glue_r_empty_tsv(output_columns))
}
# Keep annotation row rows only where both the id and name are present and non-empty.
keep <-
!is.na(annotation_meta[[id_column]]) &
nzchar(annotation_meta[[id_column]]) &
!is.na(annotation_meta[[name_column]]) &
nzchar(annotation_meta[[name_column]])
if (!any(keep)) {
return(workflow_glue_r_empty_tsv(output_columns))
}
annotation_meta <- annotation_meta[keep, , drop = FALSE]
annotation_meta <- annotation_meta[!duplicated(annotation_meta[[id_column]]), ]
out <- data.frame(
annotation_meta[[id_column]],
annotation_meta[[name_column]],
row.names = NULL,
stringsAsFactors = FALSE
)
names(out) <- output_columns
out
}
#' Extract `GENEID->gene_name` and `TXNAME->transcript_name` mappings.
#'
#' @param annotation_path Path to a GTF or GFF file.
#'
#' @return A list with `gene` and `transcript` data.frames.
workflow_glue_r_annotation_name_maps <- function(annotation_path) {
if (bambu_missing(annotation_path)) {
return(list(
gene = workflow_glue_r_empty_tsv(c("GENEID", "gene_name")),
transcript = workflow_glue_r_empty_tsv(c("TXNAME", "transcript_name"))
))
}
annotation <- rtracklayer::import(annotation_path)
annotation_meta <- S4Vectors::mcols(annotation)
list(
gene = workflow_glue_r_annotation_name_map_from_meta(
annotation_meta, "gene_id", "gene_name", "GENEID"
),
transcript = workflow_glue_r_annotation_name_map_from_meta(
annotation_meta, "transcript_id", "transcript_name", "TXNAME"
)
)
}

13
bin/workflow_glue_r/R/de_analysis.R
View File

@ -672,11 +672,13 @@ de_run_dexseq_result <- function(
run_inner(covariates)
}
de_dge_columns <- c("GENEID", "baseMean", "log2FoldChange", "lfcSE", "stat", "pvalue", "padj")
de_dge_columns <- c("GENEID", "gene_name", "baseMean", "log2FoldChange", "lfcSE", "stat", "pvalue", "padj")
de_dtu_transcript_columns <- c(
"featureID",
"groupID",
"gene_name",
"transcript_name",
"log2FoldChange",
"pvalue",
"padj",
@ -1143,6 +1145,15 @@ main_run_de_analysis <- function(args) {
} else {
dex_df <- postprocess_run$dex_df
tx_dtu <- postprocess_run$tx_dtu
if ("featureID" %in% names(tx_dtu)) {
tx_match <- match(as.character(tx_dtu$featureID), as.character(tx_meta$TXNAME))
if ("gene_name" %in% names(tx_meta)) {
tx_dtu$gene_name <- tx_meta$gene_name[tx_match]
}
if ("transcript_name" %in% names(tx_meta)) {
tx_dtu$transcript_name <- tx_meta$transcript_name[tx_match]
}
}
gene_dtu <- postprocess_run$gene_dtu
contrast_qc$dtu_status <- "SUCCESS"

103
bin/workflow_glue_r/tests/testthat/test_bambu.R
View File

@ -768,10 +768,59 @@ testthat::test_that("list columns flattened for TSV output", {
testthat::expect_equal(normalised$list_col, c("x;y", "z"))
})
testthat::test_that("annotation name maps are extracted from GTF", {
fixture_dir <- tempfile("gene-name-map-")
dir.create(fixture_dir)
gtf <- file.path(fixture_dir, "annotation.gtf")
writeLines(
c(
paste(
"chr1", "sim", "transcript", "1", "100", ".", "+", ".",
'gene_id "gene1"; transcript_id "tx1"; gene_name "GENEA"; transcript_name "TXA";',
sep = "\t"
),
paste(
"chr1", "sim", "exon", "1", "100", ".", "+", ".",
'gene_id "gene1"; transcript_id "tx1"; gene_name "GENEA"; transcript_name "TXA";',
sep = "\t"
),
paste(
"chr1", "sim", "transcript", "201", "300", ".", "+", ".",
'gene_id "gene2"; transcript_id "tx2"; gene_name "GENEB"; transcript_name "TXB";',
sep = "\t"
)
),
gtf
)
maps <- workflow_glue_r_annotation_name_maps(gtf)
gene_name_map <- maps$gene
transcript_name_map <- maps$transcript
testthat::expect_equal(names(gene_name_map), c("GENEID", "gene_name"))
testthat::expect_equal(nrow(gene_name_map), 2)
testthat::expect_equal(gene_name_map$gene_name[match("gene1", gene_name_map$GENEID)], "GENEA")
testthat::expect_equal(gene_name_map$gene_name[match("gene2", gene_name_map$GENEID)], "GENEB")
testthat::expect_equal(names(transcript_name_map), c("TXNAME", "transcript_name"))
testthat::expect_equal(nrow(transcript_name_map), 2)
testthat::expect_equal(
transcript_name_map$transcript_name[match("tx1", transcript_name_map$TXNAME)],
"TXA"
)
testthat::expect_equal(
transcript_name_map$transcript_name[match("tx2", transcript_name_map$TXNAME)],
"TXB"
)
})
# Verify all expected output files are created with correct structure.
testthat::test_that("bambu outputs written correctly", {
out_dir <- tempfile("bambu-write-")
dir.create(out_dir)
fixture_dir <- tempfile("bambu-write-fixture-")
dir.create(fixture_dir)
sample_names <- c("sampleA", "sampleB")
base_se <- make_test_tx_se(sample_names = sample_names)
@ -782,7 +831,29 @@ testthat::test_that("bambu outputs written correctly", {
)
gene_se <- make_test_gene_se(sample_names = sample_names)
sample_df <- data.frame(alias = sample_names, stringsAsFactors = FALSE)
args <- list(out_dir = out_dir, transcriptome_mode = "discover", ndr = 0.15)
annotation <- file.path(fixture_dir, "gene_names.gtf")
writeLines(
c(
paste(
"chr1", "test", "transcript", "1", "50", ".", "+", ".",
'gene_id "gene1"; transcript_id "tx1"; gene_name "GENEA"; transcript_name "TXA";',
sep = "\t"
),
paste(
"chr1", "test", "transcript", "201", "250", ".", "+", ".",
'gene_id "gene2"; transcript_id "tx3"; gene_name "GENEB"; transcript_name "TXC";',
sep = "\t"
)
),
annotation
)
args <- list(
out_dir = out_dir,
transcriptome_mode = "discover",
ndr = 0.15,
annotation = annotation
)
qc_stats <- list(
samples = 2,
total_transcripts_before_filter = 4,
@ -828,7 +899,37 @@ testthat::test_that("bambu outputs written correctly", {
testthat::expect_true("eqClassById" %in% names(tx_meta))
testthat::expect_equal(tx_meta$eqClassById[[1]], "1;2")
testthat::expect_true("gene_name" %in% names(tx_meta))
testthat::expect_true("transcript_name" %in% names(tx_meta))
testthat::expect_equal(
unique(stats::na.omit(tx_meta$gene_name[tx_meta$GENEID == "gene1"])),
"GENEA"
)
testthat::expect_equal(
tx_meta$transcript_name[match("tx1", tx_meta$TXNAME)],
"TXA"
)
testthat::expect_true(all(c("TXNAME", "sampleA", "sampleB") %in% names(tx_counts)))
tx_rds <- readRDS(file.path(out_dir, "bambu_transcripts.rds"))
gene_rds <- readRDS(file.path(out_dir, "bambu_genes.rds"))
tx_rds_meta <- as.data.frame(SummarizedExperiment::rowData(tx_rds))
gene_rds_meta <- as.data.frame(SummarizedExperiment::rowData(gene_rds))
testthat::expect_true("gene_name" %in% names(tx_rds_meta))
testthat::expect_true("transcript_name" %in% names(tx_rds_meta))
testthat::expect_true("gene_name" %in% names(gene_rds_meta))
testthat::expect_equal(
unique(stats::na.omit(tx_rds_meta$gene_name[tx_rds_meta$GENEID == "gene2"])),
"GENEB"
)
testthat::expect_equal(
tx_rds_meta$transcript_name[match("tx3", tx_rds_meta$TXNAME)],
"TXC"
)
testthat::expect_equal(
gene_rds_meta$gene_name[match("gene1", gene_rds_meta$GENEID)],
"GENEA"
)
})
testthat::test_that("collate combines multiple chunk quantification outputs", {

20
bin/workflow_glue_r/tests/testthat/test_de_analysis.R
View File

@ -813,6 +813,19 @@ testthat::test_that("CLI integration produces expected outputs", {
fixture_dir <- tempfile("de-cli-")
dir.create(fixture_dir)
bundle <- write_de_fixture_bundle(fixture_dir, levels = c("control", "treated"))
tx_se <- readRDS(bundle$transcript_rds)
tx_row_data <- as.data.frame(SummarizedExperiment::rowData(tx_se))
tx_row_data$gene_name <- c("GENEA", "GENEA", "GENEB", "GENEB")
tx_row_data$transcript_name <- c("TXA", "TXB", "TXC", "TXD")
SummarizedExperiment::rowData(tx_se) <- S4Vectors::DataFrame(tx_row_data)
saveRDS(tx_se, bundle$transcript_rds)
gene_se <- readRDS(bundle$gene_rds)
gene_row_data <- as.data.frame(SummarizedExperiment::rowData(gene_se))
gene_row_data$gene_name <- c("GENEA", "GENEB")
SummarizedExperiment::rowData(gene_se) <- S4Vectors::DataFrame(gene_row_data)
saveRDS(gene_se, bundle$gene_rds)
out_dir <- file.path(fixture_dir, "out")
result <- run_rscript(
@ -850,8 +863,11 @@ testthat::test_that("CLI integration produces expected outputs", {
)
testthat::expect_gt(nrow(dge), 0)
testthat::expect_true(all(c("GENEID", "log2FoldChange", "padj") %in% names(dge)))
testthat::expect_true(all(c("featureID", "groupID", "padj") %in% names(dtu_tx)))
testthat::expect_true(all(c("GENEID", "gene_name", "log2FoldChange", "padj") %in% names(dge)))
testthat::expect_true(all(c("featureID", "groupID", "gene_name", "transcript_name", "padj") %in% names(dtu_tx)))
testthat::expect_true(all(stats::na.omit(dge$gene_name) %in% c("GENEA", "GENEB")))
testthat::expect_true(all(stats::na.omit(dtu_tx$gene_name) %in% c("GENEA", "GENEB")))
testthat::expect_true(all(stats::na.omit(dtu_tx$transcript_name) %in% c("TXA", "TXB", "TXC", "TXD")))
testthat::expect_true(all(c("featureID", "groupID", "padj") %in% names(dexseq)))
testthat::expect_true("analysis_fallbacks" %in% names(de_qc))
testthat::expect_true("contrasts" %in% names(de_qc))

2
modules/local/bambu_chunked.nf
View File

@ -88,6 +88,7 @@ process collateBambuQuant {
memory "16 GB"
input:
tuple val(meta), path(chunk_dirs, stageAs: "chunks/*")
path annotation, stageAs: "annotation/*"
output:
tuple val(meta), path("${meta.alias}"), emit: dir
tuple val(meta), path("${meta.alias}/transcripts.gtf"), emit: gtf
@ -103,6 +104,7 @@ process collateBambuQuant {
"""
supeRglue bambu collate \
${chunk_dirs_arg} \
--annotation "${annotation}" \
--transcriptome_mode "${params.transcriptome_mode}" \
${ndr_arg} \
--out_dir "${meta.alias}"

6
subworkflows/transcriptome.nf
View File

@ -252,7 +252,8 @@ workflow transcriptome_analysis {
.groupTuple()
.map { alias, metas, chunk_dirs ->
tuple(metas[0], chunk_dirs)
}
},
analysis_annotation
)
joint_bambu_empty = runJointBambuEmpty(
bambu_empty_inputs(joint_quant_inputs_all)
@ -294,7 +295,8 @@ workflow transcriptome_analysis {
.groupTuple()
.map { alias, metas, chunk_dirs ->
tuple(metas[0], chunk_dirs)
}
},
analysis_annotation
)
sample_bambu_empty = runPerSampleBambuEmpty(
bambu_empty_inputs(sample_quant_inputs_all)