nf-core · jfy133 · Feb 5, 2025 · Jan 31, 2025 · Feb 4, 2025 · Feb 4, 2025
diff --git a/modules/nf-core/malt/build/main.nf b/modules/nf-core/malt/build/main.nf
@@ -9,6 +9,7 @@ process MALT_BUILD {
     path fastas, stageAs: 'fa_folder/'
     path gff
     path mapping_db
+    val map_type
 
     output:
     path "malt_index/", emit: index
@@ -21,16 +22,17 @@ process MALT_BUILD {
     script:
     def args = task.ext.args ?: ''
     def igff = gff ? "-igff ${gff}" : ""
-
     """
+    INDEX=`find -L . -name '*.db' -o -name '*.abin' -type f`
+    echo \$INDEX
     malt-build \\
         ${args} \\
         -v \\
-        --input fa_folder \\
+        --input fa_folder/ \\
         ${igff} \\
         -d 'malt_index/' \\
         -t ${task.cpus} \\
-        -mdb ${mapping_db}/*.db |&tee malt-build.log
+        -${map_type} \$INDEX |&tee malt-build.log
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/nf-core/malt/build/meta.yml b/modules/nf-core/malt/build/meta.yml
@@ -31,8 +31,12 @@ input:
         pattern: "*/|*.gff|*.gff3"
   - - mapping_db:
         type: file
-        description: MEGAN .db file from https://software-ab.cs.uni-tuebingen.de/download/megan6/welcome.html
-        pattern: "*.db"
+        description: An uncompressed MEGAN mapping file from https://software-ab.cs.uni-tuebingen.de/download/megan6/welcome.html (either mapping db, or acc2* .abin file)
+        pattern: "*.{db,abin}"
+  - - map_type:
+        type: string
+        description: The type of map file provided to the pipeline. Must be a valid string corresponding to a single-dash parameter, for example '-a2tax' or '-a2interpro2go'
+        pattern: "mdb|a2t|s2t|a2ec|s2ec|t4ec|a2eggnog|s2eggnog|t4eggnog|a2gtdb|s2gtdb|t4gtdb|a2interpro2go|s2interpro2go|t4interprotogo|a2kegg|s2kegg|t4kegg|a2pgpt|s2pgpt|t4pgpt|a2seed|s2seed|t4seed"
 output:
   - index:
       - malt_index/:

diff --git a/modules/nf-core/malt/build/tests/main.nf.test b/modules/nf-core/malt/build/tests/main.nf.test
@@ -10,19 +10,74 @@ nextflow_process {
     tag "malt/build"
     tag "unzip"
 
-    setup {
-        run("UNZIP") {
-            script "../../../unzip/main.nf"
+    test("sarscov2 - fastq - mdb") {
+        config "./nextflow.config"
+
+        setup {
+            run("UNZIP", alias: "UNZIP_MDB") {
+                script "../../../unzip/main.nf"
+                process {
+                    """
+                    input[0] = [[], file("https://ngi-igenomes.s3.eu-west-1.amazonaws.com/test-data/createtaxdb/taxonomy/megan-nucl-Feb2022.db.zip", checkIfExists: true)]
+                    """
+                }
+            }
+        }
+
+        when {
+            params {
+                module_args = '-J-Xmx6G --sequenceType DNA'
+            }
             process {
                 """
-                input[0] = [[], file("https://ngi-igenomes.s3.eu-west-1.amazonaws.com/test-data/createtaxdb/taxonomy/megan-nucl-Feb2022.db.zip", checkIfExists: true)]
+                def genome_1 = file(params.modules_testdata_base_path + 'genomics/prokaryotes/metagenome/fasta/sarscov2.fasta', checkIfExists: true )
+                def genome_2 = file(params.modules_testdata_base_path + 'genomics/sarscov2/genome/genome.fasta', checkIfExists: true )
+
+                input[0] = [genome_1,genome_2]
+                input[1] = []
+                input[2] = UNZIP_MDB.out.unzipped_archive.map { it[1] }
+                input[3] = 'mdb'
                 """
             }
         }
+
+        then {
+
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                        path("${process.out.index[0]}/index0.idx"),
+                        path("${process.out.index[0]}/ref.db"),
+                        path("${process.out.index[0]}/ref.idx"),
+                        path("${process.out.index[0]}/ref.inf"),
+                        path("${process.out.index[0]}/taxonomy.idx"),
+                        path("${process.out.index[0]}/taxonomy.map"),
+                        path("${process.out.index[0]}/taxonomy.tre"),
+                        process.out.versions
+                        )
+                    .match()
+                },
+                { assert path(process.out.log[0]).readLines().last().contains("Peak memory") },
+                { assert path("${process.out.index[0]}/table0.db").exists() },
+                { assert path("${process.out.index[0]}/table0.idx").exists() },
+            )
+        }
     }
 
-    test("sarscov2 - fastq") {
+    test("sarscov2 - fastq - abin") {
         config "./nextflow.config"
+
+        setup {
+            run("UNZIP", alias: "UNZIP_ABIN") {
+                script "../../../unzip/main.nf"
+                process {
+                    """
+                    input[0] = [[], file("https://software-ab.cs.uni-tuebingen.de/download/megan6/nucl_acc2tax-Jul2019.abin.zip", checkIfExists: true)]
+                    """
+                }
+            }
+        }
+
         when {
             params {
                 module_args = '-J-Xmx6G --sequenceType DNA'
@@ -34,7 +89,8 @@ nextflow_process {
 
                 input[0] = [genome_1,genome_2]
                 input[1] = []
-                input[2] = UNZIP.out.unzipped_archive.map { it[1] }
+                input[2] = UNZIP_ABIN.out.unzipped_archive.map { it[1] }
+                input[3] = 'a2t'
                 """
             }
         }
@@ -74,7 +130,8 @@ nextflow_process {
 
                 input[0] = [genome_1,genome_2]
                 input[1] = []
-                input[2] = UNZIP.out.unzipped_archive.map { it[1] }
+                input[2] = []
+                input[3] = 'mdb'
                 """
             }
         }

diff --git a/modules/nf-core/malt/build/tests/main.nf.test.snap b/modules/nf-core/malt/build/tests/main.nf.test.snap
@@ -1,4 +1,42 @@
 {
+    "sarscov2 - fastq - mdb": {
+        "content": [
+            "index0.idx:md5,0ba3d8bfb7ef28d08e2a005dd3405c55",
+            "ref.db:md5,6b36ae031c49feaae50f4cea07d9c7f4",
+            "ref.idx:md5,8ba66cdf65181c7efee1d366574cb9d7",
+            "ref.inf:md5,042712533a0187b6566db67c6503a71e",
+            "taxonomy.idx:md5,e7ce35e6238f39fa0c236fcf991546e4",
+            "taxonomy.map:md5,5bb3f2192e925bca2e61e4b54f1671e0",
+            "taxonomy.tre:md5,f76fb2d5aa9b0d637234d48175841e0e",
+            [
+                "versions.yml:md5,4ad582e415ed27dd4a275a149209961b"
+            ]
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.4"
+        },
+        "timestamp": "2025-02-04T10:27:03.034359655"
+    },
+    "sarscov2 - fastq - abin": {
+        "content": [
+            "index0.idx:md5,0ba3d8bfb7ef28d08e2a005dd3405c55",
+            "ref.db:md5,6b0b40df6094a735fa2843e0ef916a02",
+            "ref.idx:md5,42b0bc56c1562594c7500f2f86a3ca71",
+            "ref.inf:md5,042712533a0187b6566db67c6503a71e",
+            "taxonomy.idx:md5,e7ce35e6238f39fa0c236fcf991546e4",
+            "taxonomy.map:md5,5bb3f2192e925bca2e61e4b54f1671e0",
+            "taxonomy.tre:md5,f76fb2d5aa9b0d637234d48175841e0e",
+            [
+                "versions.yml:md5,4ad582e415ed27dd4a275a149209961b"
+            ]
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.4"
+        },
+        "timestamp": "2025-02-04T10:30:59.462853239"
+    },
     "sarscov2 - fastq - stub": {
         "content": [
             {
@@ -31,24 +69,5 @@
             "nextflow": "24.10.0"
         },
         "timestamp": "2024-11-15T12:14:24.942518"
-    },
-    "sarscov2 - fastq": {
-        "content": [
-            "index0.idx:md5,0ba3d8bfb7ef28d08e2a005dd3405c55",
-            "ref.db:md5,6b36ae031c49feaae50f4cea07d9c7f4",
-            "ref.idx:md5,8ba66cdf65181c7efee1d366574cb9d7",
-            "ref.inf:md5,042712533a0187b6566db67c6503a71e",
-            "taxonomy.idx:md5,e7ce35e6238f39fa0c236fcf991546e4",
-            "taxonomy.map:md5,5bb3f2192e925bca2e61e4b54f1671e0",
-            "taxonomy.tre:md5,f76fb2d5aa9b0d637234d48175841e0e",
-            [
-                "versions.yml:md5,4ad582e415ed27dd4a275a149209961b"
-            ]
-        ],
-        "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "24.10.0"
-        },
-        "timestamp": "2024-11-15T12:13:59.078459"
     }
 }
diff --git a/modules/nf-core/malt/build/tests/nextflow.config b/modules/nf-core/malt/build/tests/nextflow.config
@@ -1,4 +1,8 @@
 process {
+    withName: UNZIP {
+        ext.prefix = 'mapping_file'
+    }
+
     withName: MALT_BUILD {
         ext.args = params.module_args
     }

diff --git a/modules/nf-core/malt/run/tests/main.nf.test b/modules/nf-core/malt/run/tests/main.nf.test
@@ -15,6 +15,7 @@ nextflow_process {
         setup {
         run("UNZIP") {
             script "../../../unzip/main.nf"
+            config './nextflow.config'
             process {
                 """
                 input[0] = [[], file("https://ngi-igenomes.s3.eu-west-1.amazonaws.com/test-data/createtaxdb/taxonomy/megan-nucl-Feb2022.db.zip", checkIfExists: true)]
@@ -23,11 +24,13 @@ nextflow_process {
         }
         run("MALT_BUILD") {
             script "../../../malt/build/main.nf"
+            config './nextflow.config'
             process {
                 """
                 input[0] = file(params.modules_testdata_base_path + "genomics/sarscov2/genome/genome.fasta", checkIfExists: true)
                 input[1] = []
                 input[2] = UNZIP.out.unzipped_archive.map { it[1] }
+                input[3] = 'mdb'
                 """
             }
         }

diff --git a/modules/nf-core/malt/run/tests/nextflow.config b/modules/nf-core/malt/run/tests/nextflow.config
@@ -1,4 +1,7 @@
 process {
+    withName: UNZIP {
+        ext.prefix = 'mapping_file'
+    }
 
     withName: MALT_BUILD {
         ext.args = { "-J-Xmx${task.memory.toGiga()}G --sequenceType DNA" }