Release 1.7.9

CHANGELOG.md

@@ -1,5 +1,17 @@
 # CHANGELOG
 
+## 1.7.9: 2025-01-22
+- GCGI-1461: Fix output paths in calls to get_logger
+- GCGI-1462: Extend GSICAPBENCH report to TAR/PWGS
+- GCGI-1481: Fix raw coverage auto-population to exclude normal samples before selection in TAR assay
+- GCGI-1478: More informative logger name for plugin/helper/merger components
+- GCGI-1479: New `--pre-populate` option in `djerba.py` setup mode
+- GCGI-1413: Remove failed report plugin and allow summary plugin to handle failed reports
+- GCGI-1482: Updated total genome segment length constant in percent genome altered calculation
+- GCGI-1480: Updated CGI manager name and email
+- GCGI-1490: Remove input paths from pwgs.analysis results
+- GCGI-1492: Remove the Sequenza CNV plugin
+
 ## 1.7.8: 2024-12-12
 - GCGI-1464: Standalone script to diff two Djerba JSON reports
 - GCGI-1454: Added OncoKB definitions to WGTS40X and WGS40X assays

setup.py

@@ -23,7 +23,8 @@
     'resources/*',
     'R/*',
     'r/*',
-    'Rscripts/*'
+    'Rscripts/*',
+    'templates/*'
 ]
 
 with open("README.md", "r") as fh:
@@ -49,7 +50,10 @@
             'data/20240315-allCuratedGenes.tsv',
             'data/OncoTree.json',
             'data/NCCN_annotations.txt',
-            'data/benchmark_config.ini',
+            'data/benchmark_pwgs.ini',
+            'data/benchmark_tar.ini',
+            'data/benchmark_wgs.ini',
+            'data/benchmark_wgts.ini',
             'data/benchmark_params.json',
             'data/cytoBand.txt',
             'data/ensemble_conversion_hg38.txt',

src/bin/benchmark.py

@@ -6,7 +6,7 @@
 import sys
 
 sys.path.pop(0) # do not import from script directory
-from djerba.util.benchmark import benchmarker
+from djerba.util.benchmark_tools import benchmarker
 
 def get_parser():
     """Construct the parser for command-line arguments"""
@@ -21,7 +21,7 @@ def get_parser():
     # operations
     parser.add_argument('-i', '--input-dir', metavar='DIR', required=True, help='Directory to scan for workflow outputs, eg. ./GSICAPBENCHyymmdd/seqware-results/')
     parser.add_argument('-o', '--output-dir', metavar='DIR', required=True, help='Directory in which to generate HTML output')
-    parser.add_argument('-r', '--ref-path', metavar='FILE', required=True, help='Path to JSON file listing reference reports')
+    parser.add_argument('-r', '--ref-dir', metavar='DIR', required=True, help='Directory with reference index and reports')
     parser.add_argument('-s', '--sample', metavar='NAME', action='append', help='Sample names for directory scan; may be supplied more than once')
     parser.add_argument('-w', '--work-dir', metavar='DIR', required=True, help='Working directory in which to generate Djerba reports')
     cache_group = parser.add_mutually_exclusive_group()

src/bin/diff_reports.py

@@ -7,7 +7,7 @@
 
 sys.path.pop(0) # do not import from script directory
 
-from djerba.util.benchmark import report_equivalence_tester
+from djerba.util.benchmark_tools import report_equivalence_tester
 from djerba.util.logger import logger
 from djerba.util.validator import path_validator
 

src/bin/djerba.py

@@ -6,7 +6,7 @@
 import sys
 
 sys.path.pop(0) # do not import from script directory
-from djerba.core.main import main, arg_processor, DjerbaSubcommandError
+from djerba.core.main import main, arg_processor, DjerbaInvalidNameError
 from djerba.version import get_djerba_version
 import djerba.util.constants as constants
 
@@ -23,9 +23,10 @@ def get_parser():
     parser.add_argument('--version', action='store_true', help='Print the version number and exit')
     subparsers = parser.add_subparsers(title='subcommands', help='sub-command help', dest='subparser_name')
     setup_parser = subparsers.add_parser(constants.SETUP, help='setup for a Djerba report')
-    setup_parser.add_argument('-a', '--assay', metavar='NAME', required=True, choices=['WGTS', 'WGS', 'TAR', 'PWGS'], help='Name of assay')
+    setup_parser.add_argument('-a', '--assay', metavar='NAME', required=True, help='Name of assay (case-insensitive)')
     setup_parser.add_argument('-i', '--ini', metavar='PATH', help='Output path for INI file; defaults to config.ini in current directory')
     setup_parser.add_argument('-c', '--compact', action='store_true', help="Output required manual parameters only")
+    setup_parser.add_argument('-p', '--pre-populate', metavar='PATH', help='INI file with key/value pairs to pre-populate config')
     config_parser = subparsers.add_parser(constants.CONFIGURE, help='get configuration parameters')
     config_parser.add_argument('-i', '--ini', metavar='PATH', required=True, help='INI config file with user inputs')
     config_parser.add_argument('-o', '--ini-out', metavar='PATH', required=True, help='Path for output of fully specified INI config file')
@@ -69,7 +70,8 @@ def get_parser():
         sys.exit(0)
     try:
         ap = arg_processor(args)
-    except DjerbaSubcommandError as err:
+        main(ap.get_work_dir(), ap.get_log_level(), ap.get_log_path()).run(args)
+    except DjerbaInvalidNameError as err:
         print("{0}".format(err), file=sys.stderr)
         sys.exit(1)
-    main(ap.get_work_dir(), ap.get_log_level(), ap.get_log_path()).run(args)
+

src/lib/djerba/core/configure.py

@@ -43,7 +43,8 @@ def __init__(self, **kwargs):
         self.module_dir = kwargs[cc.MODULE_DIR]
         self.log_level = kwargs[cc.LOG_LEVEL]
         self.log_path = kwargs[cc.LOG_PATH]
-        self.logger = self.get_logger(self.log_level, __name__, self.log_path)
+        logger_name = 'djerba:'+self.identifier
+        self.logger = self.get_logger(self.log_level, logger_name, self.log_path)
         self.ini_required = set() # names of INI parameters the user must supply
         self.ini_defaults = {} # names and default values for other INI parameters
 
@@ -373,7 +374,8 @@ def __init__(self, config, identifier, log_level=logging.WARNING, log_path=None)
         # identifier is the component identifier, used to retrieve INI params
         self.config = config
         self.identifier = identifier
-        self.logger = self.get_logger(log_level, __name__, log_path)
+        logger_name = 'djerba:'+self.identifier+':config_wrapper'
+        self.logger = self.get_logger(log_level, logger_name, log_path)
 
     def get_config(self):
         return self.config

src/lib/djerba/core/html/clinical_header.html

@@ -27,11 +27,11 @@
   </tr>
   <tr>
     <td>Main contact:</td>
-    <td>Alexander Fortuna, MSc</td>
+    <td>Beatriz Lujan Toro, MSc</td>
   </tr>
   <tr>
-    <td>Phone:</td>
-    <td>416-673-8539</td>
+    <td>Email:</td>
+    <td>[email protected]</td>
   </tr>
   <tr>
     <td>Hours of Operation:</td>

src/lib/djerba/core/main.py

@@ -437,9 +437,10 @@ def run(self, args):
             assay = ap.get_assay()
             compact = ap.get_compact()
             ini_path = ap.get_ini_path()
+            pre_populate = ap.get_pre_populate()
             if ini_path == None:
                 ini_path = os.path.join(os.getcwd(), 'config.ini')
-            self.setup(assay, ini_path, compact)
+            self.setup(assay, ini_path, compact, pre_populate)
         elif mode == constants.CONFIGURE:
             ini_path = ap.get_ini_path()
             ini_path_out = ap.get_ini_out_path() # may be None
@@ -485,9 +486,9 @@ def run(self, args):
             self.logger.error(msg)
             raise RuntimeError(msg)
 
-    def setup(self, assay, ini_path, compact):
-        if assay == 'WGTS':
-            component_list = [
+    def setup(self, assay, ini_path, compact, pre_populate=None):
+        components_by_assay = {
+            'WGTS': [
                 'core',
                 'input_params_helper',
                 'provenance_helper',
@@ -504,9 +505,8 @@ def setup(self, assay, ini_path, compact):
                 'fusion',
                 'gene_information_merger',
                 'supplement.body',
-            ]
-        elif assay == 'WGS':
-            component_list = [
+            ],
+            'WGS': [
                 'core',
                 'input_params_helper',
                 'provenance_helper',
@@ -521,9 +521,8 @@ def setup(self, assay, ini_path, compact):
                 'wgts.cnv_purple',
                 'gene_information_merger',
                 'supplement.body',
-            ]
-        elif assay == 'TAR':
-            component_list = [
+            ],
+            'TAR': [
                 'core',
                 'tar_input_params_helper',
                 'provenance_helper',
@@ -537,9 +536,8 @@ def setup(self, assay, ini_path, compact):
                 'tar.swgs',
                 'gene_information_merger',
                 'supplement.body',
-            ]
-        elif assay == 'PWGS':
-            component_list = [
+            ],
+            'PWGS': [
                 'core',
                 'report_title',
                 'patient_info',
@@ -551,12 +549,20 @@ def setup(self, assay, ini_path, compact):
                 'pwgs.analysis',  
                 'supplement.body'
             ]
+        }
+        assay = assay.upper()
+        if assay in components_by_assay:
+            component_list = components_by_assay[assay]
         else:
-            msg = "Invalid assay name '{0}'".format(assay)
+            names = sorted(list(components_by_assay.keys()))
+            msg = "Invalid assay name '{0}'. ".format(assay)+\
+                "Assay names are not case-sensitive; valid names are {0}".format(names)
             self.logger.error(msg)
-            raise ValueError(msg)
+            raise DjerbaInvalidNameError(msg)
         generator = ini_generator(self.log_level, self.log_path)
         generator.write_config(component_list, ini_path, compact)
+        if pre_populate is not None:
+            self.write_pre_population(ini_path, pre_populate)
         self.logger.info("Wrote config for {0} to {1}".format(assay, ini_path))
 
     def update(self, config_path, json_path, out_dir, archive, pdf, summary_only, force):
@@ -570,17 +576,22 @@ def update(self, config_path, json_path, out_dir, archive, pdf, summary_only, fo
         # 1. INI config with core + plugins to update
         # 2. Text file to update summary only
         # The 'summary_only' argument controls which one is used
+        with open(json_path, encoding=cc.TEXT_ENCODING) as in_file:
+            data = json.loads(in_file.read())
         if summary_only:
+            # get failed/not-failed status from input data
+            failed = data[cc.PLUGINS]['summary'][cc.RESULTS]['failed']
+            failed_opt = 'true' if failed else 'false'
+            self.logger.debug('Found report failure status: '+failed_opt)
             # make an appropriate ConfigParser on-the-fly
             config_in = ConfigParser()
             config_in.add_section(cc.CORE)
             config_in.add_section('summary')
             config_in.set('summary', 'summary_file', config_path)
+            config_in.set('summary', 'failed', failed_opt)
             config = self.configure_from_parser(config_in)
         else:
             config = self.configure(config_path)
-        with open(json_path, encoding=cc.TEXT_ENCODING) as in_file:
-            data = json.loads(in_file.read())
         data_new = self.base_extract(config)
         data = self.update_data_from_file(data_new, json_path, force)
         if archive:
@@ -616,6 +627,24 @@ def upload_archive(self, data):
         else:
             self.logger.warning(f"Archiving was NOT successful: {report_id}")
 
+    def write_pre_population(self, config_path, prepop_path):
+        # write pre-population values from one INI into another
+        # sections in prepop_path not present in config_path are silently ignored
+        cp_config = ConfigParser()
+        cp_prepop = ConfigParser()
+        cp_config.read(config_path)
+        cp_prepop.read(prepop_path)
+        total = 0
+        for section in cp_config.sections():
+            if section in cp_prepop.sections():
+                for option in cp_prepop.options(section):
+                    cp_config.set(section, option, cp_prepop.get(section, option))
+                    total += 1
+        with open(config_path, 'w') as config_file:
+            cp_config.write(config_file)
+        template = "Pre-populated {0} value(s) in {1} from {2}"
+        self.logger.debug(template.format(total, config_path, prepop_path))
+
 
 class arg_processor(arg_processor_base):
     # class to process command-line args for creating a main object
@@ -632,6 +661,9 @@ def get_ini_path(self):
     def get_ini_out_path(self):
         return self._get_arg('ini_out')
 
+    def get_pre_populate(self):
+        return self._get_arg('pre_populate')
+
     def get_summary_path(self):
         return self._get_arg('summary')
 
@@ -667,6 +699,8 @@ def validate_args(self, args):
         if args.subparser_name == constants.SETUP:
             if args.ini!=None:
                 v.validate_output_file(args.ini)
+            if args.pre_populate!=None:
+                v.validate_input_file(args.pre_populate)
         elif args.subparser_name == constants.CONFIGURE:
             v.validate_input_file(args.ini)
             v.validate_output_file(args.ini_out)
@@ -695,16 +729,16 @@ def validate_args(self, args):
                 v.validate_output_dir(args.work_dir)
         elif args.subparser_name == None:
             msg = "No subcommand name given; run with -h/--help for valid names"
-            raise DjerbaSubcommandError(msg)
+            raise DjerbaInvalidNameError(msg)
         else:
             # shouldn't happen, but handle this case for completeness
-            raise DjerbaSubcommandError("Unknown subcommand: " + args.subparser_name)
+            raise DjerbaInvalidNameError("Unknown subcommand: " + args.subparser_name)
         self.logger.info("Command-line path validation finished.")
 
 class DjerbaDependencyError(Exception):
     pass
 
-class DjerbaSubcommandError(Exception):
+class DjerbaInvalidNameError(Exception):
     pass
 
 class DjerbaUpdateKeyError(Exception):