Add cpu module tests for boltz workflow

bin/msa_manager.py

@@ -249,21 +249,31 @@ def pop_match(match_fn):
     if template_yaml:
         template_order, template_entities = parse_template_yaml(template_yaml)
         for key in template_order:
-            # Prefer exact id/type match first
+            template_entity = template_entities.get(key)
+            template_type = None
+            template_payload = None
+
+            if template_entity is not None:
+                template_type = str(template_entity["seq_type"]).strip().lower()
+                template_payload = template_entity.get("seq_value")
+                if template_payload is not None:
+                    template_payload = str(template_payload).strip()
+
+            # Prefer exact id/type match first, but reject reshuffled ids whose sequence payload does not match the template
+            def payload_matches_template(entry):
+                return template_payload is None or entry["payload"] == template_payload
+
             matched = pop_match(
                 lambda entry: make_entity_key(entry["seq_type"], entry["seq_details"].get("id")) == key
+                and payload_matches_template(entry)
             )
 
             # Fallback: MMseqs may reshuffle/swap ids; match by type + sequence payload
-            if matched is None and key in template_entities:
-                template_type = str(template_entities[key]["seq_type"]).strip().lower()
-                template_payload = template_entities[key].get("seq_value")
-                if template_payload is not None:
-                    template_payload = str(template_payload).strip()
-                    matched = pop_match(
-                        lambda entry: str(entry["seq_type"]).strip().lower() == template_type
-                        and entry["payload"] == template_payload
-                    )
+            if matched is None and template_payload is not None:
+                matched = pop_match(
+                    lambda entry: str(entry["seq_type"]).strip().lower() == template_type
+                    and entry["payload"] == template_payload
+                )
 
             # Keep template entries even if no MMseqs match, so metadata is retained.
             ordered_entries.append(matched)

conf/test_nostub.config

@@ -0,0 +1,9 @@
+stubRun = false
+
+process {
+    resourceLimits = [
+        cpus: 4,
+        memory: '15.GB',
+        time: '1.h'
+    ]
+}

modules/local/boltz_fasta/tests/main.nf.test

@@ -0,0 +1,33 @@
+nextflow_process {
+
+    name "Test Process BOLTZ_FASTA"
+    script "../main.nf"
+    process "BOLTZ_FASTA"
+
+    tag "modules"
+    tag "modules_local"
+    tag "boltz_fasta"
+    profile "test_nostub"
+
+    test("converts mixed fasta to boltz yaml") {
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id: 'mixed_sample' ],
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/untagged_complex.fasta', checkIfExists: true)
+                ]
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+}

modules/local/boltz_fasta/tests/main.nf.test.snap

@@ -0,0 +1,35 @@
+{
+    "converts mixed fasta to boltz yaml": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "mixed_sample"
+                        },
+                        "mixed_sample.yaml:md5,52bdd0470032c2e6c93b886b50906c09"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,0d82efa297b86ccb1ae14852c66f16ab"
+                ],
+                "boltz_yaml": [
+                    [
+                        {
+                            "id": "mixed_sample"
+                        },
+                        "mixed_sample.yaml:md5,52bdd0470032c2e6c93b886b50906c09"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,0d82efa297b86ccb1ae14852c66f16ab"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "25.10.4"
+        },
+        "timestamp": "2026-05-27T18:09:17.307872399"
+    }
+}

modules/local/boltz_yaml_to_colabfold_fasta/tests/main.nf.test

@@ -0,0 +1,33 @@
+nextflow_process {
+
+    name "Test Process BOLTZ_YAML_TO_COLABFOLD_FASTA"
+    script "../main.nf"
+    process "BOLTZ_YAML_TO_COLABFOLD_FASTA"
+
+    tag "modules"
+    tag "modules_local"
+    tag "boltz_yaml_to_colabfold_fasta"
+    profile "test_nostub"
+
+    test("converts boltz yaml to colabfold fasta") {
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id: 'mixed_sample' ],
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/boltz/mixed.yaml', checkIfExists: true)
+                ]
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+}

modules/local/boltz_yaml_to_colabfold_fasta/tests/main.nf.test.snap

@@ -0,0 +1,35 @@
+{
+    "converts boltz yaml to colabfold fasta": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "mixed_sample"
+                        },
+                        "mixed_sample.fasta:md5,475da65322d4589b4bfe1c6374a5b1a2"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,7f72de6c3c7727358d4638fd60c2a674"
+                ],
+                "query_fasta": [
+                    [
+                        {
+                            "id": "mixed_sample"
+                        },
+                        "mixed_sample.fasta:md5,475da65322d4589b4bfe1c6374a5b1a2"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,7f72de6c3c7727358d4638fd60c2a674"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "25.10.4"
+        },
+        "timestamp": "2026-05-27T18:44:11.575259411"
+    }
+}

modules/local/mmseqs_colabfoldsearch/tests/main.nf.test

@@ -0,0 +1,42 @@
+nextflow_process {
+
+    name "Test Process MMSEQS_COLABFOLDSEARCH"
+    script "../main.nf"
+    process "MMSEQS_COLABFOLDSEARCH"
+    config "./nextflow.config"
+
+    tag "modules"
+    tag "modules_local"
+    tag "mmseqs_colabfoldsearch"
+    profile "test_nostub"
+
+    test("runs mmseqs colabfoldsearch") {
+
+        when {
+            process {
+                """
+                def dbDir = file('db')
+                dbDir.mkdirs()
+                file('db/dummy.db').text = 'x'
+
+                def uniref30Dir = file('s3://proteinfold-dataset/test-data/mini_dbs/colabfold_uniref30/*', checkIfExists: true)
+
+                input[0] = [
+                    [ id: 'mixed' ],
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/mmseqs/ubq.fasta', checkIfExists: true)
+                ]
+                input[1] = file('db/*')
+                input[2] = uniref30Dir
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+}

modules/local/mmseqs_colabfoldsearch/tests/main.nf.test.snap

@@ -0,0 +1,51 @@
+{
+    "runs mmseqs colabfoldsearch": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "mixed"
+                        },
+                        "ubq.a3m:md5,48af8d06679fa8d083df615c1bbdafb4"
+                    ]
+                ],
+                "1": [
+                    [
+                        {
+                            "id": "mixed"
+                        },
+                        "ubq.json:md5,1d038ed924926e239647485dc3219dae"
+                    ]
+                ],
+                "2": [
+                    "versions.yml:md5,4655c25b6d38ecaeb0960ba8d95c2a35"
+                ],
+                "a3m": [
+                    [
+                        {
+                            "id": "mixed"
+                        },
+                        "ubq.a3m:md5,48af8d06679fa8d083df615c1bbdafb4"
+                    ]
+                ],
+                "json": [
+                    [
+                        {
+                            "id": "mixed"
+                        },
+                        "ubq.json:md5,1d038ed924926e239647485dc3219dae"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,4655c25b6d38ecaeb0960ba8d95c2a35"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "25.10.4"
+        },
+        "timestamp": "2026-05-30T13:54:01.899878117"
+    }
+}

modules/local/mmseqs_colabfoldsearch/tests/nextflow.config

@@ -0,0 +1,6 @@
+process {
+    withName: MMSEQS_COLABFOLDSEARCH {
+        //ext.args = '--use-env=0 --pair-mode=unpaired' //https://github.com/sokrypton/ColabFold/issues/821
+        ext.args = '--use-env=0' //TODO: multimer test when above bug fixed
+    }
+}

modules/local/split_msa/tests/main.nf.test

@@ -0,0 +1,55 @@
+nextflow_process {
+
+    name "Test Process SPLIT_MSA"
+    script "../main.nf"
+    process "SPLIT_MSA"
+
+    tag "modules"
+    tag "modules_local"
+    tag "split_msa"
+    profile "test_nostub"
+
+    test("splits msa from af3 json using yaml template") {
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id: 'mixed_sample' ],
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/boltz/mixed.json', checkIfExists: true),
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/boltz/mixed.yaml', checkIfExists: true)
+                ]
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+    test("splits msa from af3 json with renamed chains using yaml template") {
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id: 'renamed_sample' ],
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/boltz/renamed.json', checkIfExists: true),
+                    file(params.pipelines_testdata_base_path + 'testdata/sequences/boltz/original.yaml', checkIfExists: true)
+                ]
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+}