Add generic hamronize.py CLI parser that can parse any tool

- Reorganise the repo to make it pip installable directly from git/conda - Gets rid of now redundant parsers - Updates test.sh to new parser format
pha4ge · Sep 25, 2020 · 0bf0538 · 0bf0538
1 parent 34e8b8f
commit 0bf0538
Show file tree

Hide file tree

Showing 39 changed files with 143 additions and 180 deletions.
diff --git a/README.md b/README.md
@@ -15,6 +15,12 @@ cd hAMRonization
 pip install -e .
 ```
 
+## Usage
+
+```
+python hamronize.py abricate ../test/data/raw_outputs/abricate/report.tsv --reference_database_version db_v_1 --analysis_software_version tool_v_1 --format json
+```
+
 ## Parsers
 
 Parsers needing tested (both automated and just sanity checking output), see [test.sh](parsers/test.sh) for example invocations.

diff --git a/hAMRonization/hAMRonization/AbricateIO.py → hAMRonization/AbricateIO.py b/hAMRonization/hAMRonization/AbricateIO.py → hAMRonization/AbricateIO.py
diff --git a/...nization/hAMRonization/AmrFinderPlusIO.py → hAMRonization/AmrFinderPlusIO.py b/...nization/hAMRonization/AmrFinderPlusIO.py → hAMRonization/AmrFinderPlusIO.py
diff --git a/hAMRonization/hAMRonization/AmrPlusPlusIO.py → hAMRonization/AmrPlusPlusIO.py b/hAMRonization/hAMRonization/AmrPlusPlusIO.py → hAMRonization/AmrPlusPlusIO.py
diff --git a/hAMRonization/hAMRonization/AribaIO.py → hAMRonization/AribaIO.py b/hAMRonization/hAMRonization/AribaIO.py → hAMRonization/AribaIO.py
diff --git a/hAMRonization/hAMRonization/CSStarIO.py → hAMRonization/CSStarIO.py b/hAMRonization/hAMRonization/CSStarIO.py → hAMRonization/CSStarIO.py
diff --git a/hAMRonization/hAMRonization/DeepArgIO.py → hAMRonization/DeepArgIO.py b/hAMRonization/hAMRonization/DeepArgIO.py → hAMRonization/DeepArgIO.py
diff --git a/hAMRonization/hAMRonization/GrootIO.py → hAMRonization/GrootIO.py b/hAMRonization/hAMRonization/GrootIO.py → hAMRonization/GrootIO.py
diff --git a/hAMRonization/hAMRonization/Interfaces.py → hAMRonization/Interfaces.py b/hAMRonization/hAMRonization/Interfaces.py → hAMRonization/Interfaces.py
@@ -131,22 +131,73 @@ def write(self, output_location=None, output_format='tsv'):
 
 
 def cli_parser(analysis_tool):
-    parser = argparse.ArgumentParser(description=f"hAMRonization parser for {analysis_tool}")
-    parser.add_argument("report", help="Path to tool report")
-    parser.add_argument("--format", default="tsv", help="Output format (tsv or json)")
-    parser.add_argument("--output", default=None, help="Output location")
+    """
+    Generate a command line hAMRonization tool for a specific tool
+    """
+    parser = generic_parser(analysis_tool)
+    args = parser.parse_args()
+    metadata = {field: getattr(args, field) for field in required_mandatory_metadata}
+
+    # parse report and write to specified
+    parsed_report = hAMRonization.parse(args.report, metadata, analysis_tool)
+    parsed_report.write(output_location=args.output,
+                        output_format=args.format)
+
+def generate_tool_subparser(subparser, analysis_tool):
+    """
+    Build the argument parser for a specific tool (used to generate a tool-specific
+    cli-parser and a generic tool parser)
+    """
+    report_file = hAMRonization._ReportFileToUse[analysis_tool]
+    tool_parser = subparser.add_parser(analysis_tool,
+                         description=f"Applies hAMRonization specification to output from {analysis_tool} ({report_file})",
+                         usage=f"hamronize.py {analysis_tool} <options>",
+                         help=f"hAMRonize {analysis_tool}'s output report i.e., {report_file}")
+
+    tool_parser.add_argument("report", help="Path to tool report")
+    tool_parser.add_argument("--format", default="tsv", help="Output format (tsv or json)")
+    tool_parser.add_argument("--output", default=None, help="Output location")
 
     # any missing mandatory fields need supplied as CLI argument
     required_mandatory_metadata = hAMRonization._RequiredToolMetadata[analysis_tool]
     for field in required_mandatory_metadata:
-        parser.add_argument(f"--{field}", required=True,
+        tool_parser.add_argument(f"--{field}", required=True,
                             help="Input string containing the "
                                 f"{field} "
                                 f"for {analysis_tool}")
+    return subparser
+
+
+def generic_cli_interface():
+    """
+    Generate a generic tool report parser that passes to the tool specific
+    parser
+    """
+    parser = argparse.ArgumentParser(description="Convert AMR gene detection tool output to hAMRonization specification format",
+                                                 prog='hamronize',
+                                                 usage='hamronize.py <tool> <options>')
+
+    parser.add_argument('-v', '--version', action='version',
+                        version=f"%(prog)s {hAMRonization.__version__}")
+
+
+    subparser = parser.add_subparsers(title="Tools with hAMRonizable reports",
+                                       help='', dest='analysis_tool')
+
+    for analysis_tool in hAMRonization._RequiredToolMetadata.keys():
+        subparser = generate_tool_subparser(subparser, analysis_tool)
+
     args = parser.parse_args()
-    metadata = {field: getattr(args, field) for field in required_mandatory_metadata}
 
-    # parse report and write to specified
-    parsed_report = hAMRonization.parse(args.report, metadata, analysis_tool)
-    parsed_report.write(output_location=args.output,
-                        output_format=args.format)
+    if args.analysis_tool:
+        required_mandatory_metadata = hAMRonization._RequiredToolMetadata[args.analysis_tool]
+        metadata = {field: getattr(args, field) for field in required_mandatory_metadata}
+
+        # parse report and write to specified
+        parsed_report = hAMRonization.parse(args.report, metadata, args.analysis_tool)
+        parsed_report.write(output_location=args.output,
+                            output_format=args.format)
+    else:
+        parser.print_help()
+        exit(1)
+
diff --git a/...ization/hAMRonization/KmerResistanceIO.py → hAMRonization/KmerResistanceIO.py b/...ization/hAMRonization/KmerResistanceIO.py → hAMRonization/KmerResistanceIO.py
diff --git a/hAMRonization/README.md b/hAMRonization/README.md
diff --git a/hAMRonization/hAMRonization/ResFamsIO.py → hAMRonization/ResFamsIO.py b/hAMRonization/hAMRonization/ResFamsIO.py → hAMRonization/ResFamsIO.py
diff --git a/hAMRonization/hAMRonization/ResFinderIO.py → hAMRonization/ResFinderIO.py b/hAMRonization/hAMRonization/ResFinderIO.py → hAMRonization/ResFinderIO.py
diff --git a/hAMRonization/hAMRonization/RgiIO.py → hAMRonization/RgiIO.py b/hAMRonization/hAMRonization/RgiIO.py → hAMRonization/RgiIO.py
diff --git a/hAMRonization/hAMRonization/SraxIO.py → hAMRonization/SraxIO.py b/hAMRonization/hAMRonization/SraxIO.py → hAMRonization/SraxIO.py
diff --git a/hAMRonization/hAMRonization/Srst2IO.py → hAMRonization/Srst2IO.py b/hAMRonization/hAMRonization/Srst2IO.py → hAMRonization/Srst2IO.py
diff --git a/hAMRonization/hAMRonization/StarAmrIO.py → hAMRonization/StarAmrIO.py b/hAMRonization/hAMRonization/StarAmrIO.py → hAMRonization/StarAmrIO.py
diff --git a/hAMRonization/hAMRonization/__init__.py → hAMRonization/__init__.py b/hAMRonization/hAMRonization/__init__.py → hAMRonization/__init__.py
@@ -1,5 +1,7 @@
 #!/usr/bin/env python
 
+__version__ = "1.0.0"
+
 from hAMRonization.hAMRonizedResult import hAMRonizedResult
 
 from hAMRonization import AbricateIO
@@ -35,6 +37,24 @@
     "resfams": ResFamsIO.ResFamsIterator
     }
 
+_ReportFileToUse = {
+    "abricate": "OUTPUT.tsv",
+    "amrfinderplus": "OUTPUT.tsv",
+    "ariba": "OUTDIR/OUTPUT.tsv",
+    "rgi": "OUTPUT.txt or OUTPUT_bwtoutput.gene_mapping_data.txt",
+    "resfinder": "data_resfinder.json",
+    "srax": "sraX_detected_ARGs.tsv",
+    "deeparg": "OUTDIR/OUTPUT.mapping.ARG",
+    "kmerresistance": "OUTPUT.KmerRes",
+    "srst2": "OUTPUT_srst2_report.tsv",
+    "groot": "OUTPUT.tsv (from `groot report`)",
+    "staramr": "resfinder.tsv",
+    "csstar": "OUTPUT.tsv",
+    "amrplusplus": "gene.tsv",
+    "resfams": "resfams.tblout"
+    }
+
+
 _RequiredToolMetadata = {
     "abricate": AbricateIO.required_metadata,
     "amrfinderplus": AmrFinderPlusIO.required_metadata,

diff --git a/...ization/hAMRonization/hAMRonizedResult.py → hAMRonization/hAMRonizedResult.py b/...ization/hAMRonization/hAMRonizedResult.py → hAMRonization/hAMRonizedResult.py
diff --git a/hamronize.py b/hamronize.py
@@ -0,0 +1,8 @@
+#!/usr/bin/env python
+
+import argparse
+import hAMRonization
+
+if __name__ == '__main__':
+
+    hAMRonization.Interfaces.generic_cli_interface()
diff --git a/parsers/abricate_report_parser.py b/parsers/abricate_report_parser.py
diff --git a/parsers/amrfinderplus_report_parser.py b/parsers/amrfinderplus_report_parser.py
diff --git a/parsers/amrplusplus_report_parser.py b/parsers/amrplusplus_report_parser.py
diff --git a/parsers/ariba_report_parser.py b/parsers/ariba_report_parser.py
diff --git a/parsers/csstar_report_parser.py b/parsers/csstar_report_parser.py
diff --git a/parsers/deeparg_report_parser.py b/parsers/deeparg_report_parser.py
diff --git a/parsers/groot_report_parser.py b/parsers/groot_report_parser.py
diff --git a/parsers/hAMRonized_parser.py b/parsers/hAMRonized_parser.py
diff --git a/parsers/kmerresistance_report_parser.py b/parsers/kmerresistance_report_parser.py
diff --git a/parsers/resfams_report_parser.py b/parsers/resfams_report_parser.py
diff --git a/parsers/resfinder_report_parser.py b/parsers/resfinder_report_parser.py
diff --git a/parsers/rgi_report_parser.py b/parsers/rgi_report_parser.py
diff --git a/parsers/srax_report_parser.py b/parsers/srax_report_parser.py
diff --git a/parsers/srst2_report_parser.py b/parsers/srst2_report_parser.py
diff --git a/parsers/staramr_report_parser.py b/parsers/staramr_report_parser.py
diff --git a/parsers/test.sh b/parsers/test.sh
diff --git a/hAMRonization/requirements.txt → requirements.txt b/hAMRonization/requirements.txt → requirements.txt
diff --git a/hAMRonization/setup.py → setup.py b/hAMRonization/setup.py → setup.py