Upgrade to v1.2.23

Added to CLI:
1. SV matrix generator
2. CNV matrix generator

Additional changes:
1. Added Dockerfile
2. Modified reference_paths to accept a volume (closes #171)

Dockerfile

@@ -0,0 +1,17 @@
+# Use an official Python runtime as a parent image
+FROM python:3.8-slim
+
+# Set the working directory in the container
+WORKDIR /usr/src/app
+
+# Install SigProfilerMatrixGenerator from PyPI
+RUN pip install SigProfilerMatrixGenerator==1.2.23
+
+# Create a non-root user named 'spm_user'
+RUN useradd -m -s /bin/bash spm_user
+
+# Change the ownership of the /usr/src/app directory and its contents to the new non-root user
+RUN chown -R spm_user:spm_user /usr/src/app
+
+# Switch to the non-root user for subsequent commands and when running the container
+USER spm_user

SigProfilerMatrixGenerator/controllers/cli_controller.py

@@ -4,6 +4,8 @@
 from SigProfilerMatrixGenerator import test_helpers
 from SigProfilerMatrixGenerator.scripts import (
     SigProfilerMatrixGeneratorFunc as mg,
+    SVMatrixGenerator as sv_mg,
+    CNVMatrixGenerator as cnv_mg,
     reference_genome_manager,
 )
 
@@ -129,6 +131,54 @@ def parse_arguments_matrix_generator(args: List[str]) -> argparse.Namespace:
     return result
 
 
+def parse_arguments_sv_matrix_generator(args: List[str]) -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Generate a structural variant (SV) matrix from input data."
+    )
+
+    # Mandatory arguments
+    parser.add_argument("input_dir", help="The directory containing the input files.")
+    parser.add_argument("project", help="The name of the project.")
+    parser.add_argument(
+        "output_dir", help="The directory where the output matrix will be stored."
+    )
+
+    result = parser.parse_args(args)
+    return result
+
+
+def parse_arguments_cnv_matrix_generator(args: List[str]) -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Generate a Copy Number Variation (CNV) matrix."
+    )
+
+    # Mandatory arguments
+    parser.add_argument(
+        "file_type",
+        choices=[
+            "ASCAT",
+            "ASCAT_NGS",
+            "SEQUENZA",
+            "ABSOLUTE",
+            "BATTENBERG",
+            "FACETS",
+            "PURPLE",
+            "TCGA",
+        ],
+        help="The type of the input file based on the CNV calling tool used (e.g., 'ASCAT').",
+    )
+    parser.add_argument(
+        "input_file", help="The absolute path to the multi-sample segmentation file."
+    )
+    parser.add_argument("project", help="The name of the project.")
+    parser.add_argument(
+        "output_path", help="The path where the output CNV matrix will be stored."
+    )
+
+    result = parser.parse_args(args)
+    return result
+
+
 class CliController:
     def dispatch_install(self, user_args: List[str]) -> None:
         parsed_args = parse_arguments_install(user_args)
@@ -161,3 +211,20 @@ def dispatch_matrix_generator(self, user_args: List[str]) -> None:
             cushion=parsed_args.cushion,
             volume=parsed_args.volume,
         )
+
+    def dispatch_sv_matrix_generator(self, user_args: List[str]) -> None:
+        parsed_args = parse_arguments_sv_matrix_generator(user_args)
+        sv_mg.generateSVMatrix(
+            input_dir=parsed_args.input_dir,
+            project=parsed_args.project,
+            output_dir=parsed_args.output_dir,
+        )
+
+    def dispatch_cnv_matrix_generator(self, user_args: List[str]) -> None:
+        parsed_args = parse_arguments_cnv_matrix_generator(user_args)
+        cnv_mg.generateCNVMatrix(
+            file_type=parsed_args.file_type,
+            input_file=parsed_args.input_file,
+            project=parsed_args.project,
+            output_path=parsed_args.output_path,
+        )

SigProfilerMatrixGenerator/scripts/MutationMatrixGenerator.py

@@ -85,7 +85,7 @@ def perm(n, seq):
     return permus
 
 
-def reference_paths(genome):
+def reference_paths(genome, volume=None):
     """
     Returns the path to the reference genomes installed with SigProfilerMatrixGenerator
 
@@ -95,7 +95,7 @@ def reference_paths(genome):
     Returns:
             chrom_path  -> path to the reference genome's chromosome files
     """
-    reference_dir = ref_install.reference_dir()
+    reference_dir = ref_install.reference_dir(secondary_chromosome_install_dir=volume)
     ref_dir = str(reference_dir.path)
     chrom_path = str(reference_dir.get_tsb_dir() / genome) + "/"
 

SigProfilerMatrixGenerator/scripts/SigProfilerMatrixGenerator_CLI.py

@@ -7,7 +7,9 @@
 def main_function():
     commands = {
         "install": "Install reference genome files (required to generate matrices).",
-        "matrix_generator": "Create mutational matrices for all types of somatic mutations.",
+        "matrix_generator": "Create mutational matrices for SBSs, DBSs, and INDELs.",
+        "sv_matrix_generator": "Create mutational matrices for SVs.",
+        "cnv_matrix_generator": "Create mutational matrices for CNVs.",
     }
 
     if len(sys.argv) < 2 or sys.argv[1].lower() not in commands:
@@ -28,6 +30,10 @@ def main_function():
         controller.dispatch_install(args)
     elif command == "matrix_generator":
         controller.dispatch_matrix_generator(args)
+    elif command == "sv_matrix_generator":
+        controller.dispatch_sv_matrix_generator(args)
+    elif command == "cnv_matrix_generator":
+        controller.dispatch_cnv_matrix_generator(args)
 
 
 def print_usage(commands):

setup.py

@@ -3,7 +3,7 @@
 
 from setuptools import setup
 
-VERSION = "1.2.22"
+VERSION = "1.2.23"
 
 # 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 = 'v1.2.22: CLI calls ReferenceGenomeManager to download reference genomes and not install.py'
+Update = 'v1.2.23: Add SV and CNV matrix generation to CLI. Add Dockerfile.'
 
 	"""
     fh = open(filename, "w")