modify custom/filterdifferentialtable to allow >=/<= (#7258)

* modify custom/filterdifferentialabundance to allow custom setting of >=/<= comparisons through ext.args * update tests and snapshots * simplified the cardinality declaration * modified the config for propd_basic.config * modify meta.yml * modify test for custom/filterdifferentialtable * Update meta.yml * modified padj_* into stat_* in custom/filterdifferentialtable * update abundance_differential_filter with stat_* instead of padj_*
nf-core · Jan 10, 2025 · 045f0fb · 045f0fb
1 parent f349d93
commit 045f0fb
Show file tree

Hide file tree

Showing 7 changed files with 118 additions and 74 deletions.
diff --git a/modules/nf-core/custom/filterdifferentialtable/main.nf b/modules/nf-core/custom/filterdifferentialtable/main.nf
@@ -9,10 +9,8 @@ process CUSTOM_FILTERDIFFERENTIALTABLE {
 
     input:
     tuple val(meta), path(input_file)
-    val(logFC_column)
-    val(FC_threshold)
-    val(padj_column)
-    val(padj_threshold)
+    tuple val(logfc_column), val(fc_threshold), val(fc_cardinality)
+    tuple val(stat_column), val(stat_threshold), val(stat_cardinality)
 
     output:
     tuple val(meta), path("*_filtered.tsv"), emit: filtered
@@ -41,14 +39,27 @@ process CUSTOM_FILTERDIFFERENTIALTABLE {
     table = pd.read_csv("${input_file}", sep=sep)
 
     # Calculate log2 fold change threshold
-    logFC_threshold = log2(float("${FC_threshold}"))
+    logfc_threshold = log2(float("${fc_threshold}"))
+
+    # define evaluation
+    def evaluate_condition(x, threshold, cardinality):
+        if cardinality == ">=":
+            return x >= threshold
+        elif cardinality == "<=":
+            return x <= threshold
+        elif cardinality == ">":
+            return x > threshold
+        elif cardinality == "<":
+            return x < threshold
+        else:
+            raise ValueError(f"Invalid cardinality: {cardinality}")
 
     # Apply filters
     mask = (
-        table["${logFC_column}"].notna() &
-        table["${padj_column}"].notna() &
-        (table["${logFC_column}"].abs() >= logFC_threshold) &
-        (table["${padj_column}"] <= float("${padj_threshold}"))
+        table["${logfc_column}"].notna() &
+        table["${stat_column}"].notna() &
+        table["${logfc_column}"].abs().apply(lambda x: evaluate_condition(x, logfc_threshold, "${fc_cardinality}")) &
+        table["${stat_column}"].apply(lambda x: evaluate_condition(x, float("${stat_threshold}"), "${stat_cardinality}"))
     )
     filtered_table = table[mask]
 

diff --git a/modules/nf-core/custom/filterdifferentialtable/meta.yml b/modules/nf-core/custom/filterdifferentialtable/meta.yml
@@ -4,6 +4,7 @@ keywords:
   - filter
   - differential expression
   - logFC
+  - significance statistic
   - p-value
 tools:
   - "pandas":
@@ -26,18 +27,30 @@ input:
         type: file
         description: Input differential expression table (CSV, TSV, or TXT format)
         pattern: "*.{csv,tsv,txt}"
-  - - logFC_column:
+  - - logfc_column:
         type: string
         description: Name of the column containing log fold change values
-  - - FC_threshold:
+    - fc_threshold:
         type: float
         description: Fold change threshold for filtering
-  - - padj_column:
+    - fc_cardinality:
         type: string
-        description: Name of the column containing adjusted p-values
-  - - padj_threshold:
+        description: |
+          Operator to compare the fold change values with the threshold.
+          Valid values are: ">=", "<=", ">", "<".
+  - - stat_column:
+        type: string
+        description: |
+          Name of the column containing the significance statistic values
+          (eg. adjusted p-values).
+    - stat_threshold:
         type: float
-        description: Adjusted p-value threshold for filtering
+        description: Statistic threshold for filtering
+    - stat_cardinality:
+        type: string
+        description: |
+          Operator to compare the column values with the threshold.
+          Valid values are: ">=", "<=", ">", "<".
 output:
   - filtered:
       - meta:

diff --git a/modules/nf-core/custom/filterdifferentialtable/tests/main.nf.test b/modules/nf-core/custom/filterdifferentialtable/tests/main.nf.test
@@ -15,10 +15,8 @@ nextflow_process {
             process {
                 """
                 input[0] = [ [ id:'test' ], file(params.modules_testdata_base_path + "genomics/mus_musculus/rnaseq_expression/SRP254919.salmon.merged.deseq2.results.tsv", checkIfExists: true) ]
-                input[1] = 'log2FoldChange'
-                input[2] = 2
-                input[3] = 'padj'
-                input[4] = 0.05
+                input[1] = Channel.of(['log2FoldChange', 2, '>='])
+                input[2] = Channel.of(['padj', 0.05, '<='])
                 """
             }
         }

diff --git a/subworkflows/nf-core/abundance_differential_filter/main.nf b/subworkflows/nf-core/abundance_differential_filter/main.nf
@@ -19,7 +19,7 @@ def mergeMaps(meta, meta2){
 workflow ABUNDANCE_DIFFERENTIAL_FILTER {
     take:
     // Things we may need to iterate
-    ch_input                 // [[meta_input], counts, analysis method, fc_threshold, padj_threshold]
+    ch_input                 // [[meta_input], counts, analysis method, fc_threshold, stat_threshold]
 
     // Workflow-wide things, we don't need to iterate
     ch_samplesheet           // [ meta_exp, samplesheet ]
@@ -30,7 +30,7 @@ workflow ABUNDANCE_DIFFERENTIAL_FILTER {
     main:
 
     // Set up how the channels crossed below will be used to generate channels for processing
-    def criteria = multiMapCriteria { meta_input, abundance, analysis_method, fc_threshold, padj_threshold, meta_exp, samplesheet, meta_contrasts, variable, reference, target ->
+    def criteria = multiMapCriteria { meta_input, abundance, analysis_method, fc_threshold, stat_threshold, meta_exp, samplesheet, meta_contrasts, variable, reference, target ->
         samples_and_matrix:
             meta_map = meta_input + [ 'method': analysis_method ]
             [meta_map, samplesheet, abundance]
@@ -39,7 +39,7 @@ workflow ABUNDANCE_DIFFERENTIAL_FILTER {
             [ meta_map, variable, reference, target ]
         filter_params:
             meta_map = mergeMaps(meta_contrasts, meta_input) + [ 'method': analysis_method ]
-            [meta_map, [ 'fc_threshold': fc_threshold, 'padj_threshold': padj_threshold ]]
+            [meta_map, [ 'fc_threshold': fc_threshold, 'stat_threshold': stat_threshold ]]
     }
 
     // For DIFFERENTIAL modules we need to cross the things we're iterating so we
@@ -142,24 +142,37 @@ workflow ABUNDANCE_DIFFERENTIAL_FILTER {
         .join(inputs.filter_params)
         .multiMap { meta, results, filter_meta ->
             def method_params = [
-                'deseq2': [fc_column: 'log2FoldChange', padj_column: 'padj'],
-                'limma' : [fc_column: 'logFC', padj_column: 'adj.P.Val'],
-                'propd' : [fc_column: 'lfc', padj_column: 'weighted_connectivity']
+                'deseq2': [
+                    fc_column: 'log2FoldChange', fc_cardinality: '>=',
+                    stat_column: 'padj', stat_cardinality: '<='
+                ],
+                'limma' : [
+                    fc_column: 'logFC', fc_cardinality: '>=',
+                    stat_column: 'adj.P.Val', stat_cardinality: '<='
+                ],
+                'propd' : [
+                    fc_column: 'lfc', fc_cardinality: '>=',
+                    stat_column: 'weighted_connectivity', stat_cardinality: '>='
+                ]
             ]
             filter_input: [meta + filter_meta, results]
-            fc_column: method_params[meta.method].fc_column
-            padj_column: method_params[meta.method].padj_column
-            fc_threshold: filter_meta.fc_threshold
-            padj_threshold: filter_meta.padj_threshold
+            fc_input: [
+                method_params[meta.method].fc_column,
+                filter_meta.fc_threshold,
+                method_params[meta.method].fc_cardinality
+            ]
+            stat_input: [
+                method_params[meta.method].stat_column,
+                filter_meta.stat_threshold,
+                method_params[meta.method].stat_cardinality
+            ]
         }
 
     // Filter differential results
     CUSTOM_FILTERDIFFERENTIALTABLE(
         ch_diff_filter_params.filter_input,
-        ch_diff_filter_params.fc_column,
-        ch_diff_filter_params.fc_threshold,
-        ch_diff_filter_params.padj_column,
-        ch_diff_filter_params.padj_threshold
+        ch_diff_filter_params.fc_input,
+        ch_diff_filter_params.stat_input
     )
 
     emit:

diff --git a/subworkflows/nf-core/abundance_differential_filter/meta.yml b/subworkflows/nf-core/abundance_differential_filter/meta.yml
@@ -10,13 +10,16 @@ input:
             description: Count matrix file
         - analysis_method:
             type: value
-            description: Analysis method (deseq2 or limma)
+            description: Analysis method (deseq2, limma, or propd)
         - fc_threshold:
             type: value
             description: Fold change threshold for filtering
-        - padj_threshold:
+        - stat_threshold:
             type: value
-            description: Adjusted p-value threshold for filtering
+            description: |
+              Threshold for filtering the significance statistics
+              (eg. adjusted p-values in the case of deseq2 or limma,
+              weighted connectivity in the case of propd)
   - ch_samplesheet:
       description: Channel with sample information
       structure:

diff --git a/subworkflows/nf-core/abundance_differential_filter/tests/main.nf.test b/subworkflows/nf-core/abundance_differential_filter/tests/main.nf.test
@@ -16,6 +16,7 @@ nextflow_workflow {
 
     test("deseq2 - mouse - basic") {
         config './deseq2_basic.config'
+        tag "deseq2"
 
         when {
             workflow {
@@ -45,7 +46,7 @@ nextflow_workflow {
                     file(testData.expression_test_data_dir + testData.abundance_file),
                     'deseq2',  // analysis method
                     1.5,       // FC threshold
-                    0.05       // padj threshold
+                    0.05       // stat (adjusted p-value) threshold
                 ])
 
                 input[0] = ch_input
@@ -73,6 +74,7 @@ nextflow_workflow {
 
     test("limma - basic - microarray") {
         config './limma_basic_microarray.config'
+        tag "limma"
 
         setup {
             run("UNTAR") {
@@ -128,7 +130,7 @@ nextflow_workflow {
                     file,
                     'limma',
                     1.5,       // FC threshold
-                    0.05       // padj threshold
+                    0.05       // stat (adjusted p-value) threshold
                 ]}
 
                 input[0] = ch_input
@@ -155,6 +157,7 @@ nextflow_workflow {
 
     test("limma - voom") {
         config './limma_voom.config'
+        tag "limma_voom"
 
         when {
             workflow {
@@ -184,7 +187,7 @@ nextflow_workflow {
                     file(testData.expression_test_data_dir + testData.abundance_file),
                     'limma',
                     1.5,       // FC threshold
-                    0.05       // padj threshold
+                    0.05       // stat (adjusted p-value) threshold
                 ])
 
                 input[0] = ch_input
@@ -211,6 +214,7 @@ nextflow_workflow {
 
     test("deseq2 - with transcript lengths") {
         config './deseq2_basic.config'
+        tag "deseq2_with_lengths"
 
         when {
             workflow {
@@ -244,7 +248,7 @@ nextflow_workflow {
                     file(testData.expression_test_data_dir + testData.abundance_file),
                     'deseq2',
                     1.5,       // FC threshold
-                    0.05       // padj threshold
+                    0.05       // stat (adjusted p-value) threshold
                 ])
 
                 input[0] = ch_input
@@ -302,7 +306,7 @@ nextflow_workflow {
                     file(testData.expression_test_data_dir + testData.abundance_file),
                     'propd',  // analysis method
                     1.5,      // FC threshold
-                    50        // weighted connectivity threshold
+                    100       // stat (weighted connectivity) threshold
                 ])
 
                 input[0] = ch_input
@@ -328,6 +332,7 @@ nextflow_workflow {
 
     test("deseq2 and limma - mouse - basic") {
         config './deseq2_limmavoom_basic.config'
+        tag "deseq2+limmavoom"
 
         when {
             workflow {
@@ -358,14 +363,14 @@ nextflow_workflow {
                         file(testData.expression_test_data_dir + testData.abundance_file),
                         'limma',
                         1.5,       // FC threshold
-                        0.05       // padj threshold
+                        0.05       // stat (adjusted p-value) threshold
                     ],
                     [
                         [ id:'test' ],
                         file(testData.expression_test_data_dir + testData.abundance_file),
                         'deseq2',
                         1.5,       // FC threshold
-                        0.05       // padj threshold
+                        0.05       // stat (adjusted p-value) threshold
                     ]
                 )
 
@@ -425,21 +430,21 @@ nextflow_workflow {
                         file(testData.expression_test_data_dir + testData.abundance_file),
                         'limma',
                         1.5,       // FC threshold
-                        0.05       // padj threshold
+                        0.05       // stat (adjusted p-value) threshold
                     ],
                     [
                         [ id:'test' ],
                         file(testData.expression_test_data_dir + testData.abundance_file),
                         'deseq2',
                         1.5,       // FC threshold
-                        0.05       // padj threshold
+                        0.05       // stat (adjusted p-value) threshold
                     ],
                     [
                         [ id:'test' ],
                         file(testData.expression_test_data_dir + testData.abundance_file),
                         'propd',
                         1.5,       // FC threshold
-                        50         // weighted connectivity threshold
+                        100        // stat (weighted connectivity) threshold
                     ]
                 )
 
@@ -469,6 +474,7 @@ nextflow_workflow {
 
     test("stub") {
         config './deseq2_basic.config'
+        tag "stub"
 
         options "-stub"
 
@@ -500,7 +506,7 @@ nextflow_workflow {
                     file(testData.expression_test_data_dir + testData.abundance_file),
                     'deseq2',
                     1.5,       // FC threshold
-                    0.05       // padj threshold
+                    0.05       // stat (adjusted p-value) threshold
                 ])
 
                 input[0] = ch_input