Merge pull request #457 from atrigila/move_clean_fasta

Reorganize `clean fasta` code
nf-core · Oct 2, 2024 · 4c64d73 · 4c64d73
2 parents 19a1c8b + 1f9922a
commit 4c64d73
Show file tree

Hide file tree

Showing 9 changed files with 32 additions and 30 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -42,6 +42,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [[#447]](https://github.com/nf-core/smrnaseq/pull/447) - Fix [Minor fixes and general pipeline cleanup](https://github.com/nf-core/smrnaseq/issues/400) - Update variable and processes names, update channel comments, remove unused modules and params.
 - [[#448]](https://github.com/nf-core/smrnaseq/pull/448) - Migrate local mirdeep to [nf-core mirdeep2 modules and subworkflow](https://github.com/nf-core/smrnaseq/issues/443) and generate [test profile for mirdeep2](https://github.com/nf-core/smrnaseq/issues/399).
 - [[#452]](https://github.com/nf-core/smrnaseq/pull/452) - Fix [Fix ch_bowtie_index channel structure](https://github.com/nf-core/smrnaseq/issues/451) and replace untarfiles with untar [replace untarfiles with untar](https://github.com/nf-core/smrnaseq/issues/449).
+- [[#457]](https://github.com/nf-core/smrnaseq/pull/457) - QC all input [fasta files and clean them](https://github.com/nf-core/smrnaseq/issues/455).
 
 ## v2.3.1 - 2024-04-18 - Gray Zinc Dalmation Patch
 

diff --git a/subworkflows/local/prepare_genome/main.nf b/subworkflows/local/prepare_genome/main.nf
@@ -74,6 +74,11 @@ workflow PREPARE_GENOME {
     // even if bowtie index is specified, there still needs to be a fasta.
     // without fasta, no genome analysis.
     if(val_fasta) {
+        // Clean fasta (replace non-ATCGs with Ns, remove whitespaces from headers)
+        CLEAN_FASTA ( ch_fasta )
+        ch_versions      = ch_versions.mix(CLEAN_FASTA.out.versions)
+        ch_fasta         = CLEAN_FASTA.out.output
+
         //Prepare bowtie index, unless specified
         //This needs to be done here as the index is used by GENOME_QUANT
         if(val_bowtie_index) {
@@ -94,16 +99,12 @@ workflow PREPARE_GENOME {
             }
 
         } else {
-            // Clean fasta (replace non-ATCGs with Ns, remove whitespaces from headers)
-            CLEAN_FASTA ( ch_fasta )
-            ch_versions      = ch_versions.mix(CLEAN_FASTA.out.versions)
 
             // Index FASTA with nf-core Bowtie1
             INDEX_GENOME ( CLEAN_FASTA.out.output )
             ch_versions      = ch_versions.mix(INDEX_GENOME.out.versions)
 
             // Set channels: clean fasta and its index
-            ch_fasta         = CLEAN_FASTA.out.output
             ch_bowtie_index  = INDEX_GENOME.out.index.collect()
         }
     }

diff --git a/tests/test_contamination_tech_reps.nf.test.snap b/tests/test_contamination_tech_reps.nf.test.snap
@@ -40,7 +40,7 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-30T15:53:12.532061"
+        "timestamp": "2024-10-01T20:06:04.830119436"
     },
     "mirna_quant_bam": {
         "content": [
@@ -64,10 +64,10 @@
             true
         ],
         "meta": {
-            "nf-test": "0.8.4",
+            "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-04T19:24:40.511510374"
+        "timestamp": "2024-10-01T20:06:04.974546479"
     },
     "mirna_quant_edger_qc": {
         "content": [
@@ -89,10 +89,10 @@
             true
         ],
         "meta": {
-            "nf-test": "0.8.4",
+            "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-04T19:24:40.56322797"
+        "timestamp": "2024-10-01T20:06:05.025175321"
     },
     "contaminant_filter_filter": {
         "content": [
@@ -101,10 +101,10 @@
             true
         ],
         "meta": {
-            "nf-test": "0.8.4",
+            "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-04T19:24:40.458015808"
+        "timestamp": "2024-10-01T20:06:04.920520728"
     },
     "mirna_quant_mirtop": {
         "content": [
@@ -115,6 +115,6 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-20T18:06:26.655506797"
+        "timestamp": "2024-10-01T20:06:05.070939602"
     }
-}
+}
diff --git a/tests/test_mirgenedb.nf.test.snap b/tests/test_mirgenedb.nf.test.snap
@@ -22,10 +22,10 @@
             "{BOWTIE_MAP_GENOME={bowtie=1.3.0, samtools=1.16.1}, BOWTIE_MAP_HAIRPIN={bowtie=1.3.0, samtools=1.16.1}, BOWTIE_MAP_MATURE={bowtie=1.3.0, samtools=1.16.1}, BOWTIE_MAP_SEQCLUSTER={bowtie=1.3.0, samtools=1.16.1}, FASTP={fastp=0.23.4}, FASTQC_RAW={fastqc=0.12.1}, FASTQC_TRIM={fastqc=0.12.1}, FORMAT_HAIRPIN={fastx_toolkit=0.0.14}, FORMAT_MATURE={fastx_toolkit=0.0.14}, INDEX_HAIRPIN={bowtie=1.3.0}, INDEX_MATURE={bowtie=1.3.0}, MIRDEEP2_MAPPER={mirdeep2=2.0.1}, MIRDEEP2_MIRDEEP2={mirdeep2=2.0.1}, PARSE_HAIRPIN={seqkit=2.6.1}, PARSE_MATURE={seqkit=2.6.1}, SAMTOOLS_FLAGSTAT={samtools=1.21}, SAMTOOLS_IDXSTATS={samtools=1.21}, SAMTOOLS_INDEX={samtools=1.21}, SAMTOOLS_SORT={samtools=1.21}, SAMTOOLS_STATS={samtools=1.21}, SEQCLUSTER_COLLAPSE={seqcluster=1.2.9}, SEQKIT_FQ2FA={seqkit=2.8.0}, SEQKIT_REPLACE={seqkit=2.8.0}, Workflow={nf-core/smrnaseq=v2.3.2dev}}"
         ],
         "meta": {
-            "nf-test": "0.8.4",
+            "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-26T18:15:04.369221417"
+        "timestamp": "2024-10-01T20:13:41.274631809"
     },
     "mirna_quant_bam": {
         "content": [

diff --git a/tests/test_nextflex.nf.test.snap b/tests/test_nextflex.nf.test.snap
@@ -40,7 +40,7 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-30T16:10:06.50556"
+        "timestamp": "2024-10-01T20:16:29.435481425"
     },
     "mirna_quant_bam": {
         "content": [
@@ -142,4 +142,4 @@
         },
         "timestamp": "2024-09-20T17:11:24.369706104"
     }
-}
+}
diff --git a/tests/test_skipfastp.nf.test b/tests/test_skipfastp.nf.test
@@ -19,7 +19,7 @@ nextflow_pipeline {
             assertAll(
                 { assert workflow.success },
                 { assert snapshot(UTILS.removeNextflowVersion("$outputDir")).match("software_versions") },
-                { assert workflow.trace.succeeded().size() == 63 },
+                { assert workflow.trace.succeeded().size() == 64 },
 
                 { assert snapshot(
                     path("$outputDir/mirna_quant/mirtop/joined_samples_mirtop.tsv").exists(),

diff --git a/tests/test_skipfastp.nf.test.snap b/tests/test_skipfastp.nf.test.snap
@@ -30,14 +30,14 @@
             "Clone1_N3_mature_hairpin_genome.sorted.idxstats:md5,2620288b88bba1ea3315414016c083a1",
             "Clone1_N1_mature_hairpin_genome.sorted.idxstats:md5,e0e4a95f5c21a926f7894cf1fbe3110b",
             "Clone1_N1_mature_hairpin_genome.sorted.flagstat:md5,62208acf0c7418d590b41318d7e17d67",
-            "Clone1_N3_mature_hairpin_genome.sorted.stats:md5,471571deab69bf7a6f4cb679035398d2",
-            "Clone1_N1_mature_hairpin_genome.sorted.stats:md5,cce099859e44a7e0adf1c304fdb14f4f"
+            "Clone1_N3_mature_hairpin_genome.sorted.stats:md5,1c03f98d16d2e418692d3227c8be2803",
+            "Clone1_N1_mature_hairpin_genome.sorted.stats:md5,e5aee4c4e099d5f4e82bd22f58309e2c"
         ],
         "meta": {
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-19T15:26:42.738485563"
+        "timestamp": "2024-10-01T20:19:25.61285999"
     },
     "software_versions": {
         "content": [
@@ -47,7 +47,7 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-30T16:13:51.93255"
+        "timestamp": "2024-10-01T20:19:25.507173099"
     },
     "mirna_quant_bam": {
         "content": [
@@ -140,6 +140,6 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-20T18:29:58.277371938"
+        "timestamp": "2024-10-01T20:19:25.557700049"
     }
-}
+}
diff --git a/tests/test_umi.nf.test b/tests/test_umi.nf.test
@@ -19,7 +19,7 @@ nextflow_pipeline {
             assertAll(
                 { assert workflow.success },
                 { assert snapshot(UTILS.removeNextflowVersion("$outputDir")).match("software_versions") },
-                { assert workflow.trace.succeeded().size() == 73 },
+                { assert workflow.trace.succeeded().size() == 74 },
 
                 { assert snapshot(
                     path("$outputDir/mirna_quant/bam/mature/SRX8195118_SRR11631014_mature.sorted.stats"),

diff --git a/tests/test_umi.nf.test.snap b/tests/test_umi.nf.test.snap
@@ -30,14 +30,14 @@
             "SRX8195117_SRR11631013_mature_hairpin_genome.sorted.flagstat:md5,977e88cbe62027285df73e1f7f9cd9bc",
             "SRX8195117_SRR11631013_mature_hairpin_genome.sorted.idxstats:md5,cc0413bf90252c3b3af8926fd64bc873",
             "SRX8195118_SRR11631014_mature_hairpin_genome.sorted.idxstats:md5,a4874de294706a7ead30258944ff2dad",
-            "SRX8195118_SRR11631014_mature_hairpin_genome.sorted.stats:md5,89f8dd2287a390cf347627c51b0d2ec2",
-            "SRX8195117_SRR11631013_mature_hairpin_genome.sorted.stats:md5,910348e352583b44c3f30cfd6332d846"
+            "SRX8195118_SRR11631014_mature_hairpin_genome.sorted.stats:md5,6beddbebbd48d4ffc01e059c73a5a330",
+            "SRX8195117_SRR11631013_mature_hairpin_genome.sorted.stats:md5,cc5ef7cbd5a29794170ac6cff59dfd76"
         ],
         "meta": {
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-19T15:35:36.816602685"
+        "timestamp": "2024-10-01T20:24:14.961610878"
     },
     "software_versions": {
         "content": [
@@ -47,7 +47,7 @@
             "nf-test": "0.9.0",
             "nextflow": "24.04.4"
         },
-        "timestamp": "2024-09-30T16:23:12.559357"
+        "timestamp": "2024-10-01T20:24:14.901461898"
     },
     "mirna_quant_bam": {
         "content": [
@@ -160,4 +160,4 @@
         },
         "timestamp": "2024-09-20T19:12:28.290360163"
     }
-}
+}