Merge branch 'dev' into seqtk_sample_size_warning

Author: maxulysse

CHANGELOG.md

@@ -39,10 +39,11 @@ Initial release of nf-core/seqinspector, created with the [nf-core](https://nf-c
 ### `Fixed`
 
 - [#71](https://github.com/nf-core/seqinspector/pull/71) FASTQSCREEN does not fail when multiple reads are provided.
+- [#94](https://github.com/nf-core/seqinspector/issues/94) Go through and validate test data
 - [#99](https://github.com/nf-core/seqinspector/pull/99) Fix group reports for paired reads
 - [#107](https://github.com/nf-core/seqinspector/pull/107) Put SeqFU-stats section reports together
 - [#112](https://github.com/nf-core/seqinspector/pull/112) Making fastq_screen_references value to use parentDir
-- [#94] (https://github.com/nf-core/seqinspector/issues/94) Go through and validate test data
+- [#121](https://github.com/nf-core/seqinspector/pull/121) Cleanup sample naming for MultiQC report (#105)
 - [#162](https://github.com/nf-core/seqinspector/pull/162) Fix bugs in qc_bam and prepare_genome subworkflows and add tests
 - [#163](https://github.com/nf-core/seqinspector/pull/163) Run fastqscreen with subsampled data if available
 

assets/multiqc_config.yml

@@ -15,3 +15,22 @@ report_section_order:
 export_plots: true
 
 disable_version_detection: true
+
+fn_clean_trim:
+  - "_screen" # Added by FastqScreen
+
+# Make sample name with indexes a bit prettier
+# for SE: "SampleName_01" -> "SampleName #01"
+# for PE: "SampleName_01_1" -> "SampleName #01_1"
+#    and: "SampleName_01_2" -> "SampleName #01_2"
+sample_names_replace_regex: True
+sample_names_replace:
+  "(.+)_(\\d{2,}|\\d{2,}_[12])$": "\\1 #\\2"
+
+table_sample_merge:
+  "Read1":
+    - type: regex
+      pattern: "_1$"
+  "Read2":
+    - type: regex
+      pattern: "_2$"

subworkflows/local/utils_nfcore_seqinspector_pipeline/main.nf

@@ -100,6 +100,8 @@ workflow PIPELINE_INITIALISATION {
     //
     // Create channel from input file provided through params.input
     //
+    nr_samples = Channel.fromList(samplesheetToList(params.input, "${projectDir}/assets/schema_input.json"))
+        .toList().size()
 
     channel
         .fromList(samplesheetToList(params.input, "${projectDir}/assets/schema_input.json"))
@@ -108,7 +110,9 @@ workflow PIPELINE_INITIALISATION {
         .map {
             meta, fastq_1, fastq_2, idx ->
                 def tags = meta.tags ? meta.tags.tokenize(":") : []
-                def updated_meta = meta + [ id:"${meta.sample}_${idx}", tags:tags ]
+                def pad_positions = [nr_samples.length(), 2].max()
+                def zero_padded_idx = idx.padLeft(pad_positions, "0")
+                def updated_meta = meta + [ id:"${meta.sample}_${zero_padded_idx}", tags:tags ]
                 if (!fastq_2) {
                     return [
                         updated_meta.id,

tests/MiSeq.main.nf.test.snap

@@ -17,12 +17,12 @@
             },
             [
                 "fastqc",
-                "fastqc/hg001_1_fastqc.html",
-                "fastqc/hg001_1_fastqc.zip",
+                "fastqc/hg001_01_fastqc.html",
+                "fastqc/hg001_01_fastqc.zip",
                 "fastqscreen",
-                "fastqscreen/hg001_1_screen.html",
-                "fastqscreen/hg001_1_screen.png",
-                "fastqscreen/hg001_1_screen.txt",
+                "fastqscreen/hg001_01_screen.html",
+                "fastqscreen/hg001_01_screen.png",
+                "fastqscreen/hg001_01_screen.txt",
                 "multiqc",
                 "multiqc/global_report",
                 "multiqc/global_report/multiqc_data",
@@ -47,7 +47,7 @@
                 "multiqc/global_report/multiqc_data/multiqc_fastq_screen.txt",
                 "multiqc/global_report/multiqc_data/multiqc_fastqc.txt",
                 "multiqc/global_report/multiqc_data/multiqc_general_stats.txt",
-                "multiqc/global_report/multiqc_data/multiqc_seqfu.txt",
+                "multiqc/global_report/multiqc_data/multiqc_seqfu_mqc.txt",
                 "multiqc/global_report/multiqc_data/multiqc_software_versions.txt",
                 "multiqc/global_report/multiqc_data/multiqc_sources.txt",
                 "multiqc/global_report/multiqc_plots",
@@ -66,7 +66,7 @@
                 "multiqc/global_report/multiqc_plots/pdf/fastqc_sequence_duplication_levels_plot.pdf",
                 "multiqc/global_report/multiqc_plots/pdf/fastqc_sequence_length_distribution_plot.pdf",
                 "multiqc/global_report/multiqc_plots/pdf/fastqc_top_overrepresented_sequences_table.pdf",
-                "multiqc/global_report/multiqc_plots/pdf/seqfu.pdf",
+                "multiqc/global_report/multiqc_plots/pdf/seqfu_mqc.pdf",
                 "multiqc/global_report/multiqc_plots/png",
                 "multiqc/global_report/multiqc_plots/png/fastq_screen_plot-cnt.png",
                 "multiqc/global_report/multiqc_plots/png/fastq_screen_plot-pct.png",
@@ -82,7 +82,7 @@
                 "multiqc/global_report/multiqc_plots/png/fastqc_sequence_duplication_levels_plot.png",
                 "multiqc/global_report/multiqc_plots/png/fastqc_sequence_length_distribution_plot.png",
                 "multiqc/global_report/multiqc_plots/png/fastqc_top_overrepresented_sequences_table.png",
-                "multiqc/global_report/multiqc_plots/png/seqfu.png",
+                "multiqc/global_report/multiqc_plots/png/seqfu_mqc.png",
                 "multiqc/global_report/multiqc_plots/svg",
                 "multiqc/global_report/multiqc_plots/svg/fastq_screen_plot-cnt.svg",
                 "multiqc/global_report/multiqc_plots/svg/fastq_screen_plot-pct.svg",
@@ -98,46 +98,46 @@
                 "multiqc/global_report/multiqc_plots/svg/fastqc_sequence_duplication_levels_plot.svg",
                 "multiqc/global_report/multiqc_plots/svg/fastqc_sequence_length_distribution_plot.svg",
                 "multiqc/global_report/multiqc_plots/svg/fastqc_top_overrepresented_sequences_table.svg",
-                "multiqc/global_report/multiqc_plots/svg/seqfu.svg",
+                "multiqc/global_report/multiqc_plots/svg/seqfu_mqc.svg",
                 "multiqc/global_report/multiqc_report.html",
                 "pipeline_info",
                 "pipeline_info/nf_core_seqinspector_software_mqc_versions.yml",
                 "samtools_faidx",
                 "samtools_faidx/genome.fa.fai",
                 "samtools_faidx/genome.fa.sizes",
                 "seqfu_stats",
-                "seqfu_stats/hg001_1_seqfu.tsv",
-                "seqfu_stats/hg001_1_seqfu_mqc.txt"
+                "seqfu_stats/hg001_01_seqfu.tsv",
+                "seqfu_stats/hg001_01_seqfu_mqc.txt"
             ],
             [
-                "hg001_1_screen.png:md5,b9c58e388f18ad11f4c2e29e9b573a5e",
-                "hg001_1_screen.txt:md5,8e53ea043d658a130f4b8ca455e2ba9b",
-                "fastq_screen_plot.txt:md5,3023b86477fecf6d46d8fc60d82cc53a",
-                "fastqc-status-check-heatmap.txt:md5,825668e94de27356f0a112beb758e416",
-                "fastqc_adapter_content_plot.txt:md5,1fa991080c20755e5e729a5eeab960e5",
-                "fastqc_overrepresented_sequences_plot.txt:md5,e810a3d3eba0ba83b122763505f511a7",
-                "fastqc_per_base_n_content_plot.txt:md5,cbdb64298ff839665e8c8c8e89896e87",
-                "fastqc_per_base_sequence_quality_plot.txt:md5,4962950bc3ce77169ac510d64cb769c6",
-                "fastqc_per_sequence_gc_content_plot_Counts.txt:md5,e62273533304be59dab5f13afef93b3e",
-                "fastqc_per_sequence_gc_content_plot_Percentages.txt:md5,3d49290a15406cb5dac1460c02729f0f",
-                "fastqc_per_sequence_quality_scores_plot.txt:md5,7cc10d1bc92ca09431a506c38d911228",
-                "fastqc_sequence_counts_plot.txt:md5,0425d7298c24df6158d5bf7aca1d4190",
-                "fastqc_sequence_duplication_levels_plot.txt:md5,9111d942c7eed5af0e9ccb8c42db8505",
-                "fastqc_sequence_length_distribution_plot.txt:md5,04adfcd0125ebcc3e3bb631f1956c812",
+                "hg001_01_screen.png:md5,b9c58e388f18ad11f4c2e29e9b573a5e",
+                "hg001_01_screen.txt:md5,8e53ea043d658a130f4b8ca455e2ba9b",
+                "fastq_screen_plot.txt:md5,f6883339d43a090e86e02fdf3feacbbd",
+                "fastqc-status-check-heatmap.txt:md5,0ac9450d210de9ab4548378101877c8e",
+                "fastqc_adapter_content_plot.txt:md5,6e2a760185573c44180ca43081cc7c99",
+                "fastqc_overrepresented_sequences_plot.txt:md5,251ad51e27e966476911ff4c77c2a893",
+                "fastqc_per_base_n_content_plot.txt:md5,230b2a0319f3b7490498e8edb47b17bf",
+                "fastqc_per_base_sequence_quality_plot.txt:md5,4934bce34aec037e1ea590bd69f74781",
+                "fastqc_per_sequence_gc_content_plot_Counts.txt:md5,028ada8fc060ce4af7476f339c57c50d",
+                "fastqc_per_sequence_gc_content_plot_Percentages.txt:md5,6bab830c22525c15a6048d8054f685cf",
+                "fastqc_per_sequence_quality_scores_plot.txt:md5,fd29224ed588e37e1fc85bf7f717e31e",
+                "fastqc_sequence_counts_plot.txt:md5,cc83b291e21249f3c4e25d34cc898f0f",
+                "fastqc_sequence_duplication_levels_plot.txt:md5,7b7665c78ca66d3e13b1850701b198dc",
+                "fastqc_sequence_length_distribution_plot.txt:md5,d7bac93b11316846778b9d550f323371",
                 "multiqc_citations.txt:md5,5f52d7a0141e4234c6069df9ef575c9a",
-                "multiqc_fastq_screen.txt:md5,7c5d711e27c53d13c81644a1f2468dba",
-                "multiqc_fastqc.txt:md5,1a4b472e13cadc770832b0e20d1de7b0",
-                "multiqc_seqfu.txt:md5,9595d608ff5709f2bdec8cb1016416c0",
+                "multiqc_fastq_screen.txt:md5,0ab0d1c4c3e3393afdb50ad0e116e394",
+                "multiqc_fastqc.txt:md5,b71aaffef9767dfe9d83fde0d31a3a1b",
+                "multiqc_seqfu_mqc.txt:md5,9595d608ff5709f2bdec8cb1016416c0",
                 "genome.fa.fai:md5,59557cd3a71ca5e0840285b6088ec1d6",
                 "genome.fa.sizes:md5,f154a58a25ba9c4d386003eaa1bb731b",
-                "hg001_1_seqfu.tsv:md5,b133752c897afbb4d8721b911fa117d4",
-                "hg001_1_seqfu_mqc.txt:md5,42e3506b02ceb72a6f05cb174505e722"
+                "hg001_01_seqfu.tsv:md5,b133752c897afbb4d8721b911fa117d4",
+                "hg001_01_seqfu_mqc.txt:md5,42e3506b02ceb72a6f05cb174505e722"
             ]
         ],
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.2"
         },
-        "timestamp": "2025-12-03T11:01:17.46398"
+        "timestamp": "2025-12-12T10:05:23.380990836"
     }
 }