Add cached cd hit est reference collapse step to pathoscope

Dockerfile

@@ -1,8 +1,9 @@
 FROM python:3.13-bookworm AS deps
 WORKDIR /app
-COPY --from=ghcr.io/virtool/tools:1.1.0 /tools/bowtie2/2.5.4/bowtie* /usr/local/bin/
-COPY --from=ghcr.io/virtool/tools:1.1.0 /tools/pigz/2.8/pigz /usr/local/bin/
-COPY --from=ghcr.io/virtool/tools:1.1.0 /tools/samtools/1.22.1/bin/samtools /usr/local/bin/
+COPY --from=ghcr.io/virtool/tools:1.2.0 /tools/bowtie2/2.5.4/bowtie* /usr/local/bin/
+COPY --from=ghcr.io/virtool/tools:1.2.0 /tools/cd-hit/4.8.1/cd-hit-est /usr/local/bin/
+COPY --from=ghcr.io/virtool/tools:1.2.0 /tools/pigz/2.8/pigz /usr/local/bin/
+COPY --from=ghcr.io/virtool/tools:1.2.0 /tools/samtools/1.22.1/bin/samtools /usr/local/bin/
 
 FROM python:3.13-bookworm AS uv
 WORKDIR /app

fixtures.py

@@ -2,32 +2,70 @@
 from types import SimpleNamespace
 
 from pyfixtures import fixture
+from virtool.workflow.data.index_sqlite import INDEX_SQLITE_FILE_NAME
+
+
+def get_reference_index_path(work_path: Path) -> Path:
+    return work_path / "reference_index" / "reference"
+
+
+def get_collapsed_reference_path(work_path: Path) -> Path:
+    return work_path / "collapsed_reference" / INDEX_SQLITE_FILE_NAME
+
+
+def get_subtraction_indexes_path(work_path: Path) -> Path:
+    return work_path / "subtraction_indexes"
+
+
+def get_isolate_path(work_path: Path) -> Path:
+    return work_path / "isolates"
+
+
+def get_isolate_fastq_path(isolate_path: Path) -> Path:
+    return isolate_path / "isolate_mapped.fq"
+
+
+def get_isolate_index_path(isolate_path: Path) -> Path:
+    return isolate_path / "isolates"
+
+
+def get_isolate_bam_path(isolate_path: Path) -> Path:
+    return isolate_path / "to_isolates.bam"
+
+
+def get_subtracted_bam_path(work_path: Path) -> Path:
+    return work_path / "subtracted.bam"
 
 
 @fixture
 def intermediate():
     """A namespace for storing intermediate values."""
     return SimpleNamespace(
-        to_otus=set(),
+        candidate_sequence_ids=set(),
     )
 
 
 @fixture
 def isolate_path(work_path: Path):
-    path = work_path / "isolates"
+    path = get_isolate_path(work_path)
     path.mkdir()
 
     return path
 
 
 @fixture
 def reference_index_path(work_path: Path):
-    return work_path / "reference_index" / "reference"
+    return get_reference_index_path(work_path)
+
+
+@fixture
+def collapsed_reference_path(work_path: Path):
+    return get_collapsed_reference_path(work_path)
 
 
 @fixture
 def subtraction_indexes_path(work_path: Path) -> Path:
-    return work_path / "subtraction_indexes"
+    return get_subtraction_indexes_path(work_path)
 
 
 @fixture
@@ -37,17 +75,17 @@ def isolate_fasta_path(isolate_path: Path):
 
 @fixture
 def isolate_fastq_path(isolate_path: Path):
-    return isolate_path / "isolate_mapped.fq"
+    return get_isolate_fastq_path(isolate_path)
 
 
 @fixture
 def isolate_index_path(isolate_path: Path):
-    return isolate_path / "isolates"
+    return get_isolate_index_path(isolate_path)
 
 
 @fixture
 def isolate_bam_path(isolate_path: Path):
-    return isolate_path / "to_isolates.bam"
+    return get_isolate_bam_path(isolate_path)
 
 
 @fixture
@@ -58,4 +96,4 @@ def p_score_cutoff():
 @fixture
 def subtracted_bam_path(work_path: Path) -> Path:
     """The path to the BAM file after subtraction reads have been eliminated."""
-    return work_path / "subtracted.bam"
+    return get_subtracted_bam_path(work_path)

pyproject.toml

@@ -37,7 +37,7 @@ features = ["pyo3/extension-module"]
 asyncio_mode = "auto"
 
 [tool.uv.sources]
-virtool = { git = "https://github.com/virtool/virtool", tag = "38.48.0" }
+virtool = { git = "https://github.com/virtool/virtool", tag = "39.63.0" }
 
 [build-system]
 requires = ["maturin>=1.14.1,<2.0"]