Merge pull request #27 from EBI-Metagenomics/feature/extend_metadata_…

…platform Feature/extend metadata platform
EBI-Metagenomics · Aug 30, 2024 · 0cd9a1a · 0cd9a1a
2 parents 7046b59 + a747143
commit 0cd9a1a
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 3 deletions.
diff --git a/Containerfile b/Containerfile
@@ -1,7 +1,7 @@
 FROM mambaorg/micromamba:1.5.8
 
 LABEL maintainer="Microbiome Informatics"
-LABEL version="1.0.0"
+LABEL version="1.0.1"
 LABEL description="EBI Fetch Tool."
 
 COPY --chown=$MAMBA_USER:$MAMBA_USER conda_environment.yml /tmp/env.yaml

diff --git a/conda_environment.yml b/conda_environment.yml
@@ -1,8 +1,7 @@
 name: fetchtool
 channels:
-  - bioconda
   - conda-forge
-  - defaults
+  - bioconda
 dependencies:
   - python=3.10
   - pip=24.0

diff --git a/fetchtool/fetch_reads.py b/fetchtool/fetch_reads.py
@@ -36,6 +36,7 @@ class FetchReads(AbstractDataFetcher):
         "experiment_accession",
         "run_accession",
         "instrument_model",
+        "instrument_platform",
         "library_layout",
         "fastq_ftp",
         "fastq_md5",
@@ -207,6 +208,8 @@ def _retrieve_project_info_from_api(self, project_accession):
                             "LIBRARY_STRATEGY": d.get("library_strategy"),
                             "LIBRARY_SOURCE": d.get("library_source"),
                             "LIBRARY_LAYOUT": d.get("library_layout"),
+                            "INSTRUMENT_MODEL": d.get("instrument_model"),
+                            "INSTRUMENT_PLATFORM": d.get("instrument_platform"),
                         }
                     )
         return mapped_data
@@ -228,6 +231,8 @@ def map_project_info_to_row(self, run):
             "file_path": run["DATA_FILE_PATH"],
             "library_strategy": run["LIBRARY_STRATEGY"],
             "library_source": run["LIBRARY_SOURCE"],
+            "instrument_model": run["INSTRUMENT_MODEL"],
+            "instrument_platform": run["INSTRUMENT_PLATFORM"],
         }
 
     def _get_project_accessions_from_runs(self, runs):

diff --git a/tests/unit/test_fetch_reads.py b/tests/unit/test_fetch_reads.py
@@ -91,6 +91,8 @@ def test_map_project_info_db_row_should_copy_fields(self):
             "LIBRARY_LAYOUT": "PAIRED",
             "LIBRARY_SOURCE": "METAGENOMIC",
             "LIBRARY_STRATEGY": "WGS",
+            "INSTRUMENT_MODEL": "Illumina HiSeq 2000",
+            "INSTRUMENT_PLATFORM": "ILLUMINA",
             "file": "ERR599383_1.fastq.gz;ERR599383_2.fastq.gz",
             "DATA_FILE_PATH": "/tmp/ERP001736/ERR599383_1.fastq.gz;/tmp/ERP001736/ERR599383_2.fastq.gz",
         }
@@ -103,6 +105,8 @@ def test_map_project_info_db_row_should_copy_fields(self):
             ("LIBRARY_LAYOUT", "library_layout"),
             ("LIBRARY_STRATEGY", "library_strategy"),
             ("LIBRARY_SOURCE", "library_source"),
+            ("INSTRUMENT_MODEL", "instrument_model"),
+            ("INSTRUMENT_PLATFORM", "instrument_platform"),
         )
         for f1, f2 in equivalent_fields:
             assert raw_data[f1] == transform[f2]
@@ -129,6 +133,7 @@ def mock_get_run_metadata(self, *args, **kwargs):
                 "experiment_accession": "ERX2789866",
                 "run_accession": "ERR2777788",
                 "instrument_model": "unspecified",
+                "instrument_platform": "ILLUMINA",
                 "library_layout": "PAIRED",
                 "fastq_ftp": "ftp.sra.ebi.ac.uk/vol1/fastq/ERR277/009/ERR2777788/ERR2777788_1.txt.gz;"
                 "ftp.sra.ebi.ac.uk/vol1/fastq/ERR277/009/ERR2777790/ERR2777788_2.txt.gz",
@@ -147,6 +152,7 @@ def mock_get_run_metadata(self, *args, **kwargs):
                 "experiment_accession": "ERX2789866",
                 "run_accession": "ERR2777789",
                 "instrument_model": "unspecified",
+                "instrument_platform": "LS454",
                 "library_layout": "PAIRED",
                 "fastq_ftp": "",
                 "fastq_md5": "",
@@ -164,6 +170,7 @@ def mock_get_run_metadata(self, *args, **kwargs):
                 "experiment_accession": "ERX2789867",
                 "run_accession": "ERR2777790",
                 "instrument_model": "unspecified",
+                "instrument_platform": "LS454",
                 "library_layout": "PAIRED",
                 "fastq_ftp": "ftp.sra.ebi.ac.uk/vol1/fastq/ERR277/009/ERR2777790/ERR2777790_1.fastq.gz;"
                 "ftp.sra.ebi.ac.uk/vol1/fastq/ERR277/009/ERR2777790/ERR2777790_2.fastq.gz",