Merge pull request #170 from AlexandrovLab/cli_updates

v1.2.22: CLI installs reference genomes using the ReferenceGenomeMana…

SigProfilerMatrixGenerator/controllers/cli_controller.py

@@ -1,8 +1,11 @@
 import argparse
 from typing import List
 
-from SigProfilerMatrixGenerator import install, test_helpers
-from SigProfilerMatrixGenerator.scripts import SigProfilerMatrixGeneratorFunc as mg
+from SigProfilerMatrixGenerator import test_helpers
+from SigProfilerMatrixGenerator.scripts import (
+    SigProfilerMatrixGeneratorFunc as mg,
+    reference_genome_manager,
+)
 
 
 def parse_arguments_test(args: List[str]) -> argparse.Namespace:
@@ -42,11 +45,11 @@ def parse_arguments_install(args: List[str]) -> argparse.Namespace:
     )
     parser.add_argument(
         "-l",
-        "--local_install_genome",
+        "--local_genome",
         help="""
             Install an offline reference genome downloaded from the Alexandrov Lab's FTP server.
             Provide the absolute path to the locally-stored genome file.
-            For downloads, visit AlexandrovLab's server:
+            For downloads, visit AlexandrovLab's ftp server:
             ftp://alexandrovlab-ftp.ucsd.edu/pub/tools/SigProfilerMatrixGenerator/
             """,
         default=None,
@@ -129,11 +132,13 @@ def parse_arguments_matrix_generator(args: List[str]) -> argparse.Namespace:
 class CliController:
     def dispatch_install(self, user_args: List[str]) -> None:
         parsed_args = parse_arguments_install(user_args)
-        install.install(
-            parsed_args.genome,
-            offline_files_path=parsed_args.local_install_genome,
-            volume=parsed_args.volume,
-        )
+        rgm = reference_genome_manager.ReferenceGenomeManager(parsed_args.volume)
+        # ftp genome installation (default)
+        if parsed_args.local_genome is None:
+            rgm.download_genome(parsed_args.genome)
+        # local genome installation
+        else:
+            rgm.install_local_genome(parsed_args.genome, parsed_args.local_genome)
 
     def dispatch_test(self, user_args: List[str]) -> None:
         parsed_args = parse_arguments_test(user_args)

SigProfilerMatrixGenerator/scripts/reference_genome_manager.py

@@ -7,6 +7,7 @@
 import logging
 import time
 
+from pathlib import Path
 from SigProfilerMatrixGenerator.scripts import ref_install
 
 # Constants
@@ -379,6 +380,39 @@ def download_genome(self, genome_name):
         local_filepath.unlink()
         logging.info(f"{genome_name} has been successfully installed.")
 
+    def install_local_genome(self, genome_name, local_genome_dir):
+        """
+        Install a reference genome originating from the FTP server that is stored locally.
+
+        - genome_name (str): The name of the genome.
+        - local_genome_dir (Path or str): The local directory path where the genome archive is stored.
+        """
+
+        local_genome_dir = Path(local_genome_dir)
+        archive_file_path = local_genome_dir / f"{genome_name}.tar.gz"
+
+        # Verify that the local genome file exists
+        if not archive_file_path.exists():
+            logging.error(f"Local genome file {archive_file_path} does not exist.")
+            return
+
+        # Extract the archive
+        try:
+            self._unzip_file(archive_file_path)
+        except tarfile.TarError as e:
+            logging.error(f"Error extracting the archive: {e}")
+            return
+
+        # Verify that all necessary files are extracted and have correct checksums
+        if not self.is_genome_installed(genome_name):
+            logging.error(f"Installation verification failed for {genome_name}.")
+            self.print_genome_checksum_verification_report(genome_name)
+            return
+
+        logging.info(
+            f"{genome_name} has been successfully installed from the local file."
+        )
+
     def is_genome_installed(self, genome_name):
         """
         Verifies whether all files for specified genome is fully and correctly installed.

setup.py

@@ -3,7 +3,7 @@
 
 from setuptools import setup
 
-VERSION = "1.2.21"
+VERSION = "1.2.22"
 
 # remove the dist folder first if exists
 if os.path.exists("dist"):
@@ -23,7 +23,7 @@ def write_version_py(filename="SigProfilerMatrixGenerator/version.py"):
 # THIS FILE IS GENERATED FROM SIGPROFILEMATRIXGENERATOR SETUP.PY
 short_version = '%(version)s'
 version = '%(version)s'
-Update = 'v1.2.21: Add reference_genome_manager and remove site-package benchmarking.'
+Update = 'v1.2.22: CLI calls ReferenceGenomeManager to download reference genomes and not install.py'
 
 	"""
     fh = open(filename, "w")

tests/controllers/test_cli_controller.py

@@ -2,11 +2,26 @@
 
 import pytest
 
-from SigProfilerMatrixGenerator import install, test_helpers
+from SigProfilerMatrixGenerator import test_helpers
+from SigProfilerMatrixGenerator import install
 from SigProfilerMatrixGenerator.controllers import cli_controller
+from SigProfilerMatrixGenerator.scripts import reference_genome_manager
 
 
 class TestController:
+    @pytest.fixture
+    def mock_ref_gen_manager(self, monkeypatch):
+        """Mock ReferenceGenomeManager for testing."""
+        mock_manager = mock.create_autospec(
+            reference_genome_manager.ReferenceGenomeManager, instance=True
+        )
+        monkeypatch.setattr(
+            reference_genome_manager,
+            "ReferenceGenomeManager",
+            lambda volume: mock_manager,
+        )
+        return mock_manager
+
     @pytest.fixture
     def mock_install(self, monkeypatch):
         """Temporarily change install.install with a mock object for testing"""
@@ -21,11 +36,16 @@ def mock_test(self, monkeypatch):
         monkeypatch.setattr(test_helpers, "test_one_genome", mock_test)
         return mock_test
 
-    def test_dispatch_install(self, mock_install):
+    def test_dispatch_download_ftp_genome(self, mock_ref_gen_manager):
+        controller = cli_controller.CliController()
+        controller.dispatch_install(["GRCh37"])
+        mock_ref_gen_manager.download_genome.assert_called_with("GRCh37")
+
+    def test_dispatch_install_local_genome(self, mock_ref_gen_manager):
         controller = cli_controller.CliController()
-        controller.dispatch_install(["GRCh37", "--local_install_genome", "/somewhere"])
-        mock_install.assert_called_with(
-            "GRCh37", offline_files_path="/somewhere", volume=None
+        controller.dispatch_install(["yeast", "--local_genome", "/somewhere"])
+        mock_ref_gen_manager.install_local_genome.assert_called_with(
+            "yeast", "/somewhere"
         )
 
     genome_calls = [