Merge pull request #132 from pinin4fjords/fix_excluded_samples

Fix up sample exclusion options
nf-core · Jul 25, 2023 · 28f0ae0 · 28f0ae0
2 parents 0ba8a56 + 8a91954
commit 28f0ae0
Show file tree

Hide file tree

Showing 9 changed files with 92 additions and 46 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -12,13 +12,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [[#130](https://github.com/nf-core/differentialabundance/pull/130)] - Document reasons for lack of differential expression ([@pinin4fjords](https://github.com/pinin4fjords), review by [@jfy133](https://github.com/jfy133))
 - [[#131](https://github.com/nf-core/differentialabundance/pull/131)] - Improve gtf to table configurability ([@pinin4fjords](https://github.com/pinin4fjords), review by [@WackerO](https://github.com/WackerO))
 - [[#137](https://github.com/nf-core/differentialabundance/pull/137)] - Add `--sizefactors_from_controls` and `--gene_id_col` for DESeq2 module to modules.config ([@WackerO](https://github.com/WackerO), review by [@pinin4fjords](https://github.com/pinin4fjords))
+- [[#145](https://github.com/nf-core/differentialabundance/pull/145)] - Template update for nf-core/tools v2.9 ([@nf-core-bot](https://github.com/nf-core-bot), review by [@pinin4fjords](https://github.com/pinin4fjords), [@WackerO](https://github.com/WackerO))
 
 ### `Fixed`
 
 - [[#116](https://github.com/nf-core/differentialabundance/issues/116)] - Skip outlier detection with low replication ([@pinin4fjords](https://github.com/pinin4fjords), review by [@nvnieuwk](https://github.com/nvnieuwk))
 - [[#122](https://github.com/nf-core/differentialabundance/pull/126)] - Add spaces to satisfy nf-core download for singularity ([@pinin4fjords](https://github.com/pinin4fjords), review by [@WackerO](https://github.com/WackerO))
 - [[#127](https://github.com/nf-core/differentialabundance/issues/127)] - [Bug] Can't pass samplesheet with -c file.config , or -params-file params.yml or directly with --input samplesheet.csv ([@ctuni](https://github.com/ctuni), review by [@pinin4fjords](https://github.com/pinin4fjords))
-- [[#138 ](https://github.com/nf-core/differentialabundance/issues/138)]- Fix bugs with --control_features and --sizefactors_from_controls ([@WackerO](https://github.com/WackerO), review by [@pinin4fjords](https://github.com/pinin4fjords))
+- [[#138](https://github.com/nf-core/differentialabundance/issues/138)]- Fix bugs with --control_features and --sizefactors_from_controls ([@WackerO](https://github.com/WackerO), review by [@pinin4fjords](https://github.com/pinin4fjords))
+- [[#133](https://github.com/nf-core/differentialabundance/issues/133)] - Sample exclusion options fail due to contrast-wise normalisation ([@pinin4fjords](https://github.com/pinin4fjords), review by [@WackerO](https://github.com/WackerO))
 
 ## v1.2.0 - 2023-04-19
 

diff --git a/assets/differentialabundance_report.Rmd b/assets/differentialabundance_report.Rmd
@@ -295,10 +295,7 @@ treatment-mCherry-hND6-batcheffect.deseq2.results.tsv
 
 ```{r, echo=FALSE}
 
-prefix_part_names <- c('variable', 'reference', 'target', 'blocking')
-diff_prefixes <- sub('-$', '', apply(contrasts[,prefix_part_names], 1, function(x) paste(x, collapse = '-')))
-
-differential_files <- lapply(diff_prefixes, function(d){
+differential_files <- lapply(contrasts$id, function(d){
  file.path(params$input_dir, paste0(gsub(' |;', '_', d), params$differential_file_suffix)) 
 })
 
@@ -328,16 +325,31 @@ differential_results <- lapply(differential_files, function(diff_file){
   }
   diff
 })
-names(differential_results) <- diff_prefixes
+names(differential_results) <- contrasts$id
 ```
 
 <!-- Calculate some summary statistics -->
 
 ```{r, echo=FALSE}
 
-contrast_descriptions <- paste(contrasts$target, 'versus', contrasts$reference, 'in', contrasts$variable) 
-with_blocking <- which(contrasts$blocking != '')
-contrast_descriptions[with_blocking] <- paste0(contrast_descriptions[with_blocking], " (blocking on ", contrasts$blocking[with_blocking],")")
+# Function to make friendly contrast name from contrast components, including optional bits
+
+name_contrast <- function(i){
+  contrast_name <- paste(contrasts$target[i], 'versus', contrasts$reference[i], 'in', contrasts$variable[i])
+  contrast_vals <- contrasts[i,]
+  populated <- colnames(contrasts)[! (is.na(contrast_vals) | contrast_vals == '' | is.null(contrast_vals))]
+  optional <- setdiff(populated, c('id', 'target', 'reference', 'variable'))
+
+  if (length(optional) > 0){
+    optional_part <-  paste0('(', paste(paste(optional, contrasts[i,optional], sep=': '), collapse=', '), ')')
+  }else{
+    optional_part <- ''
+  }
+
+  paste(contrast_name, optional_part)
+}
+
+contrast_descriptions <- unlist(lapply(1:nrow(contrasts), function(x) name_contrast(x)))
 
 # Check both adjusted and unadjusted p values
 

diff --git a/conf/modules.config b/conf/modules.config
@@ -100,17 +100,45 @@ process {
         ].join(' ').trim() }
     }
 
-    withName: DESEQ2_DIFFERENTIAL {
+    withName: DESEQ2_NORM {
+        ext.prefix = 'all'
         publishDir = [
             [
-                path: { "${params.outdir}/tables/differential" },
+                path: { "${params.outdir}/tables/processed_abundance" },
                 mode: params.publish_dir_mode,
-                pattern: '*.deseq2.results.tsv'
+                pattern: '*.{normalised_counts,vst,rlog}.tsv'
             ],
+        ]
+        ext.args = { [
+            "--sample_id_col \"${params.observations_id_col}\"",
+            "--test $params.deseq2_test",
+            "--fit_type $params.deseq2_fit_type",
+            "--sf_type $params.deseq2_sf_type",
+            "--min_replicates_for_replace $params.deseq2_min_replicates_for_replace",
+            "--use_t $params.deseq2_use_t",
+            "--lfc_threshold $params.deseq2_lfc_threshold",
+            "--alt_hypothesis $params.deseq2_alt_hypothesis",
+            "--independent_filtering $params.deseq2_independent_filtering",
+            "--p_adjust_method $params.deseq2_p_adjust_method",
+            "--alpha $params.deseq2_alpha",
+            "--minmu $params.deseq2_minmu",
+            "--vs_method $params.deseq2_vs_method",
+            "--vs_blind $params.deseq2_vs_blind",
+            "--vst_nsub $params.deseq2_vst_nsub",
+            "--shrink_lfc $params.deseq2_shrink_lfc",
+            "--cores $params.deseq2_cores",
+            "--subset_to_contrast_samples $params.differential_subset_to_contrast_samples",
+            ((meta.blocking == null) ? '' : "--blocking_variables $meta.blocking"),
+        ].join(' ').trim() }
+    }
+
+    withName: DESEQ2_DIFFERENTIAL {
+        ext.prefix = { "${meta.id}" }
+        publishDir = [
             [
-                path: { "${params.outdir}/tables/processed_abundance" },
+                path: { "${params.outdir}/tables/differential" },
                 mode: params.publish_dir_mode,
-                pattern: '*.{normalised_counts,vst,rlog}.tsv'
+                pattern: '*.deseq2.results.tsv'
             ],
             [
                 path: { "${params.outdir}/plots/qc" },
@@ -151,6 +179,7 @@ process {
     }
 
     withName: LIMMA_DIFFERENTIAL {
+        ext.prefix = { "${meta.id}" }
         publishDir = [
             [
                 path: { "${params.outdir}/tables/differential" },

diff --git a/modules.json b/modules.json
@@ -7,7 +7,7 @@
                 "nf-core": {
                     "affy/justrma": {
                         "branch": "master",
-                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
+                        "git_sha": "d0b4fc03af52a1cc8c6fb4493b921b57352b1dd8",
                         "installed_by": ["modules"]
                     },
                     "atlasgeneannotationmanipulation/gtf2featureannotation": {
@@ -37,7 +37,7 @@
                     },
                     "deseq2/differential": {
                         "branch": "master",
-                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
+                        "git_sha": "d0b4fc03af52a1cc8c6fb4493b921b57352b1dd8",
                         "installed_by": ["modules"]
                     },
                     "gsea/gsea": {
@@ -52,7 +52,7 @@
                     },
                     "limma/differential": {
                         "branch": "master",
-                        "git_sha": "911696ea0b62df80e900ef244d7867d177971f73",
+                        "git_sha": "d0b4fc03af52a1cc8c6fb4493b921b57352b1dd8",
                         "installed_by": ["modules"]
                     },
                     "rmarkdownnotebook": {
@@ -82,7 +82,7 @@
                     },
                     "untar": {
                         "branch": "master",
-                        "git_sha": "5c460c5a4736974abde2843294f35307ee2b0e5e",
+                        "git_sha": "d0b4fc03af52a1cc8c6fb4493b921b57352b1dd8",
                         "installed_by": ["modules"]
                     }
                 }

diff --git a/modules/nf-core/affy/justrma/main.nf b/modules/nf-core/affy/justrma/main.nf
diff --git a/modules/nf-core/deseq2/differential/templates/deseq_de.R b/modules/nf-core/deseq2/differential/templates/deseq_de.R
diff --git a/modules/nf-core/limma/differential/templates/limma_de.R b/modules/nf-core/limma/differential/templates/limma_de.R
diff --git a/modules/nf-core/untar/main.nf b/modules/nf-core/untar/main.nf
diff --git a/workflows/differentialabundance.nf b/workflows/differentialabundance.nf
@@ -87,6 +87,7 @@ include { SHINYNGS_APP                                      } from '../modules/n
 include { SHINYNGS_STATICEXPLORATORY as PLOT_EXPLORATORY    } from '../modules/nf-core/shinyngs/staticexploratory/main'
 include { SHINYNGS_STATICDIFFERENTIAL as PLOT_DIFFERENTIAL  } from '../modules/nf-core/shinyngs/staticdifferential/main'
 include { SHINYNGS_VALIDATEFOMCOMPONENTS as VALIDATOR       } from '../modules/nf-core/shinyngs/validatefomcomponents/main'
+include { DESEQ2_DIFFERENTIAL as DESEQ2_NORM                } from '../modules/nf-core/deseq2/differential/main'
 include { DESEQ2_DIFFERENTIAL                               } from '../modules/nf-core/deseq2/differential/main'
 include { LIMMA_DIFFERENTIAL                                } from '../modules/nf-core/limma/differential/main'
 include { CUSTOM_MATRIXFILTER                               } from '../modules/nf-core/custom/matrixfilter/main'
@@ -270,6 +271,14 @@ workflow DIFFERENTIALABUNDANCE {
     }
     else{
 
+	// 
+
+	DESEQ2_NORM (
+            ch_contrasts.first(),
+            ch_samples_and_matrix,
+            ch_control_features
+        )
+
         // Run the DESeq differential module, which doesn't take the feature
         // annotations
 
@@ -278,16 +287,16 @@ workflow DIFFERENTIALABUNDANCE {
             ch_samples_and_matrix,
             ch_control_features
         )
-
+        
         // Let's make the simplifying assumption that the processed matrices from
         // the DESeq runs are the same across contrasts. We run the DESeq process
         // with matrices once for each contrast because DESeqDataSetFromMatrix()
         // takes the model, and the model can vary between contrasts if the
         // blocking factors included differ. But the normalised and
         // variance-stabilised matrices are not (IIUC) impacted by the model.
 
-        ch_norm = DESEQ2_DIFFERENTIAL.out.normalised_counts.first()
-        ch_vst = DESEQ2_DIFFERENTIAL.out.vst_counts.first()
+        ch_norm = DESEQ2_NORM.out.normalised_counts
+        ch_vst = DESEQ2_NORM.out.vst_counts
         ch_differential = DESEQ2_DIFFERENTIAL.out.results
 
         ch_versions = ch_versions