SigProfilerAssignment 0.0.30__tar.gz → 0.0.31__tar.gz

{SigProfilerAssignment-0.0.30/SigProfilerAssignment.egg-info → SigProfilerAssignment-0.0.31}/PKG-INFO

@@ -1,12 +1,11 @@
 Metadata-Version: 2.1
 Name: SigProfilerAssignment
-Version: 0.0.30
+Version: 0.0.31
 Summary: Mutational signatures attribution and decomposition tool
 Home-page: https://github.com/AlexandrovLab/SigProfilerAssignment.git
 Author: Raviteja Vangara
 Author-email: rvangara@health.ucsd.edu
 License: UCSD
-Platform: UNKNOWN
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
 
@@ -175,5 +174,3 @@ This software and its documentation are copyright 2022 as a part of the SigProfi
 
 ## <a name="contact"></a> Contact Information
 Please address any queries or bug reports to Raviteja Vangara at rvangara@health.ucsd.edu or Marcos Díaz-Gay at mdiazgay@health.ucsd.edu.
-
-

{SigProfilerAssignment-0.0.30 → SigProfilerAssignment-0.0.31}/SigProfilerAssignment/decompose_subroutines.py

@@ -480,7 +480,7 @@ def signature_decomposition(signatures, mtype, directory, genome_build="GRCh37",
 def make_final_solution(processAvg, allgenomes, allsigids, layer_directory, m, index, allcolnames, process_std_error = "none", signature_stabilities = " ", \
                         signature_total_mutations= " ", signature_stats = "none",  cosmic_sigs=False, attribution= 0, denovo_exposureAvg  = "none", add_penalty=0.05, \
                         remove_penalty=0.01, initial_remove_penalty=0.05, de_novo_fit_penalty=0.02, background_sigs=0, genome_build="GRCh37", sequence="genome", export_probabilities=True, export_probabilities_per_mutation=False, \
-                        refit_denovo_signatures=True, collapse_to_SBS96=True, connected_sigs=True, pcawg_rule=False, verbose=False,make_plots = True, samples='./', input_type='matrix', denovo_refit_option=True):
+                        refit_denovo_signatures=True, collapse_to_SBS96=True, connected_sigs=True, pcawg_rule=False, verbose=False,make_plots = True, samples='./', input_type='matrix', denovo_refit_option=True, exome=False):
 
     if processAvg.shape[0]==allgenomes.shape[0] and processAvg.shape[0] != 96:
         collapse_to_SBS96=False
@@ -794,7 +794,7 @@ def make_final_solution(processAvg, allgenomes, allsigids, layer_directory, m, i
         if export_probabilities==True:
             if input_type=='vcf':
                 if m=='96' or m=='78' or m=='83':
-                    probability_per_mutation, samples_prob_per_mut = probabilities_per_mutation(probability, samples, m)
+                    probability_per_mutation, samples_prob_per_mut = probabilities_per_mutation(probability, samples, m, exome)
 
                     if denovo_refit_option==True:
                         if refit_denovo_signatures==True:
@@ -941,7 +941,7 @@ def probabilities(W, H, index, allsigids, allcolnames):
 
 
 ################################################### Generation of probabilities for each processes given to A mutation ############################################
-def probabilities_per_mutation(probability_matrix, samples_path, m):  
+def probabilities_per_mutation(probability_matrix, samples_path, m, exome=False):  
 #
     probability_matrix=probability_matrix.reset_index()
 #
@@ -963,11 +963,15 @@ def probabilities_per_mutation(probability_matrix, samples_path, m):
 #
     all_mutations = pd.DataFrame()
     for file in seqinfo_files:
-        try:
-            new = pd.read_csv(seqinfo_path + file, sep='\t',header=None)
-            all_mutations = pd.concat([all_mutations, new])
-        except (pd.errors.EmptyDataError):
-            pass
+        if 'exome' in file:
+            exome_df = pd.read_csv(seqinfo_path + file, sep='\t',header=None)
+        else:
+            try:
+                new = pd.read_csv(seqinfo_path + file, sep='\t',header=None)
+                all_mutations = pd.concat([all_mutations, new])
+            except (pd.errors.EmptyDataError):
+                pass
+
     all_mutations[3] = all_mutations[3].str[interval_low:interval_high]
     if m=='96' or m=='78':
         del all_mutations[4]
@@ -977,6 +981,14 @@ def probabilities_per_mutation(probability_matrix, samples_path, m):
         del all_mutations[4]
 
     all_mutations.columns = ['Sample Names', 'Chr', 'Pos', 'MutationType']
+    if exome==True:
+        del exome_df[2]
+        del exome_df[3]
+        del exome_df[4]
+        exome_df.columns = ['Chr', 'Pos']
+        exome_df['Chr'] = [str(x) for x in (exome_df['Chr']).to_list()]
+        all_mutations['Chr'] = [str(x) for x in (all_mutations['Chr']).to_list()]
+        all_mutations = pd.merge(all_mutations, exome_df)
 #
     all_samples_mutations = [y for x, y in all_mutations.groupby('Sample Names')]
 #

{SigProfilerAssignment-0.0.30 → SigProfilerAssignment-0.0.31}/SigProfilerAssignment/decomposition.py

@@ -412,7 +412,7 @@ def spa_analyze(samples, output, input_type='matrix', context_type="96", signatu
                                     add_penalty=nnls_add_penalty, remove_penalty=nnls_remove_penalty,
                                     initial_remove_penalty=init_rem_denovo,connected_sigs=connected_sigs,refit_denovo_signatures=False,
                                     make_plots=make_plots,export_probabilities=export_probabilities, export_probabilities_per_mutation=export_probabilities_per_mutation,
-                                    samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option)
+                                    samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option, exome=exome)
 
         else:
             signature_stabilities=devopts['signature_stabilities']
@@ -426,7 +426,7 @@ def spa_analyze(samples, output, input_type='matrix', context_type="96", signatu
                                     background_sigs=background_sigs, verbose=verbose, genome_build=genome_build, signature_total_mutations = signature_total_mutations,
                                     add_penalty=nnls_add_penalty, remove_penalty=nnls_remove_penalty, process_std_error = processSTE, signature_stabilities = signature_stabilities,
                                     initial_remove_penalty=init_rem_denovo,connected_sigs=connected_sigs,refit_denovo_signatures=True,export_probabilities=export_probabilities,
-                                    export_probabilities_per_mutation=export_probabilities_per_mutation, samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option)
+                                    export_probabilities_per_mutation=export_probabilities_per_mutation, samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option, exome=exome)
         
         if make_metadata:
             with open(os.path.join(output,"JOB_METADATA_SPA.txt"),"a") as sysdata:
@@ -541,7 +541,7 @@ def spa_analyze(samples, output, input_type='matrix', context_type="96", signatu
                                 make_plots=make_plots,
                                 export_probabilities=export_probabilities,
                                 export_probabilities_per_mutation=export_probabilities_per_mutation,
-                                samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option)
+                                samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option, exome=exome)
 
         if make_metadata:
             with open(os.path.join(output,"JOB_METADATA_SPA.txt"),"a") as sysdata:
@@ -659,7 +659,7 @@ def spa_analyze(samples, output, input_type='matrix', context_type="96", signatu
                             make_plots =make_plots,
                             export_probabilities=export_probabilities,
                             export_probabilities_per_mutation=export_probabilities_per_mutation,
-                            samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option)
+                            samples=samples, input_type=input_type, denovo_refit_option=denovo_refit_option, exome=exome)
         if make_metadata:
             with open(os.path.join(output,"JOB_METADATA_SPA.txt"),"a") as sysdata:
                 current_time_end = datetime.datetime.now()

SigProfilerAssignment-0.0.31/SigProfilerAssignment/version.py

@@ -0,0 +1,8 @@
+
+# THIS FILE IS GENERATED FROM SigProfilerAssignment SETUP.PY
+short_version = '0.0.31'
+version = '0.0.31'
+Update = 'v0.0.31: Resolve issue with export probability per mutation for exome data'
+
+    
+    

{SigProfilerAssignment-0.0.30 → SigProfilerAssignment-0.0.31/SigProfilerAssignment.egg-info}/PKG-INFO

@@ -1,12 +1,11 @@
 Metadata-Version: 2.1
 Name: SigProfilerAssignment
-Version: 0.0.30
+Version: 0.0.31
 Summary: Mutational signatures attribution and decomposition tool
 Home-page: https://github.com/AlexandrovLab/SigProfilerAssignment.git
 Author: Raviteja Vangara
 Author-email: rvangara@health.ucsd.edu
 License: UCSD
-Platform: UNKNOWN
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
 
@@ -175,5 +174,3 @@ This software and its documentation are copyright 2022 as a part of the SigProfi
 
 ## <a name="contact"></a> Contact Information
 Please address any queries or bug reports to Raviteja Vangara at rvangara@health.ucsd.edu or Marcos Díaz-Gay at mdiazgay@health.ucsd.edu.
-
-

{SigProfilerAssignment-0.0.30 → SigProfilerAssignment-0.0.31}/SigProfilerAssignment.egg-info/SOURCES.txt

@@ -14,7 +14,6 @@ SigProfilerAssignment.egg-info/dependency_links.txt
 SigProfilerAssignment.egg-info/not-zip-safe
 SigProfilerAssignment.egg-info/requires.txt
 SigProfilerAssignment.egg-info/top_level.txt
-SigProfilerAssignment/DecompositionPlots/.DS_Store
 SigProfilerAssignment/DecompositionPlots/PlotDecomposition.py
 SigProfilerAssignment/DecompositionPlots/PlotDecomposition_CNV48.py
 SigProfilerAssignment/DecompositionPlots/PlotDecomposition_DBS78.py
@@ -26,7 +25,6 @@ SigProfilerAssignment/DecompositionPlots/SigProfilerPlottingMatrix.py
 SigProfilerAssignment/DecompositionPlots/ReferenceFiles/Accolade_fermante.png
 SigProfilerAssignment/DecompositionPlots/ReferenceFiles/CREDIT.md
 SigProfilerAssignment/DecompositionPlots/ReferenceFiles/Fonts/Arial Bold.ttf
-SigProfilerAssignment/data/.DS_Store
 SigProfilerAssignment/data/21_breast_WGS_substitutions.mat
 SigProfilerAssignment/data/CNV_features.tsv
 SigProfilerAssignment/data/CNV_signatures.txt
@@ -146,7 +144,6 @@ SigProfilerAssignment/data/Reference_Signatures/rn6/COSMIC_v3_SBS_rn6.txt
 SigProfilerAssignment/data/Reference_Signatures/rn6/COSMIC_v3_SBS_rn6_exome.txt
 SigProfilerAssignment/data/tests/cnv_input/all.breast.ascat.summary.sample.tsv
 SigProfilerAssignment/data/tests/txt_input/DBS78_S3_Signatures.txt
-SigProfilerAssignment/data/tests/txt_input/Example_SBS_mutational_matrix_10samples.txt
 SigProfilerAssignment/data/tests/txt_input/ID83_S3_Signatures.txt
 SigProfilerAssignment/data/tests/txt_input/SBS96_S3_Signatures.txt
 SigProfilerAssignment/data/tests/txt_input/sample_matrix_DBS.txt

SigProfilerAssignment-0.0.31/setup.py

@@ -0,0 +1,66 @@
+from setuptools import setup
+import shutil
+import os
+
+# remove the dist folder first if exists
+if os.path.exists("dist"):
+    shutil.rmtree("dist")
+
+VERSION = "0.0.31"
+
+
+def write_version_py(filename="SigProfilerAssignment/version.py"):
+    # Copied from numpy setup.py
+    cnt = """
+# THIS FILE IS GENERATED FROM SigProfilerAssignment SETUP.PY
+short_version = '%(version)s'
+version = '%(version)s'
+Update = 'v0.0.31: Resolve issue with export probability per mutation for exome data'
+
+    
+    """
+    fh = open(filename, "w")
+    fh.write(
+        cnt
+        % {
+            "version": VERSION,
+        }
+    )
+    fh.close()
+
+
+with open("README.md") as f:
+    long_description = f.read()
+
+requirements = [
+    "scipy>=1.6.3",
+    "numpy>=1.21.2",
+    "pandas>=1.2.4,<2.0.0",
+    "SigProfilerMatrixGenerator>=1.2.14",
+    "sigProfilerPlotting>=1.3.11",
+    "statsmodels>=0.9.0",
+    "scikit-learn>=0.24.2",
+    "psutil>=5.6.1",
+    "reportlab>=3.5.42",
+    "PyPDF2>=3.0.0",
+    "alive_progress>=2.4.1",
+    "pdf2image>=1.16.0",
+    "PyMuPDF>=1.21.0",
+]
+
+write_version_py()
+setup(
+    name="SigProfilerAssignment",
+    version=VERSION,
+    description="Mutational signatures attribution and decomposition tool",
+    long_description=long_description,
+    long_description_content_type="text/markdown",  # This is important!
+    url="https://github.com/AlexandrovLab/SigProfilerAssignment.git",
+    author="Raviteja Vangara",
+    author_email="rvangara@health.ucsd.edu",
+    license="UCSD",
+    packages=["SigProfilerAssignment"],
+    install_requires=requirements,
+    include_package_data=True,
+    zip_safe=False,
+)

SigProfilerAssignment-0.0.30/SigProfilerAssignment/data/tests/txt_input/Example_SBS_mutational_matrix_10samples.txt

@@ -1,97 +0,0 @@
-MutationType	Sample_1	Sample_2	Sample_3	Sample_4	Sample_5	Sample_6	Sample_7	Sample_8	Sample_9	Sample_10
-A[C>A]A	152	88	159	33	54	96	114	66	146	12
-A[C>A]C	73	53	104	27	27	54	68	53	68	11
-A[C>A]G	13	11	21	5	5	11	14	10	13	2
-A[C>A]T	71	48	86	19	26	52	61	37	70	7
-A[C>G]A	50	51	106	29	18	51	67	56	48	11
-A[C>G]C	31	33	64	16	11	34	43	32	32	6
-A[C>G]G	11	9	18	5	4	9	12	9	10	2
-A[C>G]T	54	55	111	28	19	57	73	56	54	11
-A[C>T]A	177	148	358	114	90	138	176	198	148	49
-A[C>T]C	92	73	209	54	37	68	97	102	75	22
-A[C>T]G	253	517	355	424	592	533	175	348	378	268
-A[C>T]T	118	97	225	68	50	92	125	127	100	29
-A[T>A]A	38	30	70	20	13	29	42	40	32	8
-A[T>A]C	34	28	64	21	13	25	37	39	27	9
-A[T>A]G	46	36	85	25	16	34	50	50	38	9
-A[T>A]T	42	37	72	17	14	39	49	34	43	6
-A[T>C]A	209	142	385	121	65	119	212	250	147	45
-A[T>C]C	65	49	119	38	24	44	68	75	51	15
-A[T>C]G	173	116	317	99	53	97	175	206	121	36
-A[T>C]T	181	133	332	106	64	117	188	211	136	40
-A[T>G]A	19	24	46	11	6	26	32	21	21	4
-A[T>G]C	13	21	37	9	4	23	27	14	18	3
-A[T>G]G	39	35	78	21	12	34	49	43	35	8
-A[T>G]T	25	50	80	45	8	57	62	26	42	5
-C[C>A]A	174	62	106	20	63	69	79	40	156	7
-C[C>A]C	66	48	83	18	25	52	59	35	67	7
-C[C>A]G	38	13	32	9	13	12	19	19	30	3
-C[C>A]T	97	46	84	17	34	50	60	35	90	6
-C[C>G]A	34	32	70	19	10	32	44	38	30	7
-C[C>G]C	45	40	90	24	14	39	56	49	40	9
-C[C>G]G	18	10	23	7	6	9	13	14	14	3
-C[C>G]T	51	51	105	26	16	53	69	52	50	9
-C[C>T]A	110	87	245	53	37	85	119	107	96	22
-C[C>T]C	97	79	231	50	31	75	111	103	83	18
-C[C>T]G	219	313	304	265	344	313	151	264	257	159
-C[C>T]T	115	97	245	61	36	93	137	125	99	23
-C[T>A]A	17	10	29	9	5	8	15	19	11	3
-C[T>A]C	29	28	58	15	10	29	38	28	31	5
-C[T>A]G	28	19	46	14	10	17	27	27	23	5
-C[T>A]T	24	30	55	15	8	32	39	23	32	4
-C[T>C]A	59	52	125	35	18	49	75	72	50	13
-C[T>C]C	52	47	103	33	20	46	63	58	56	12
-C[T>C]G	95	68	179	56	30	59	101	112	75	20
-C[T>C]T	71	62	144	47	22	59	87	80	86	14
-C[T>G]A	12	13	27	8	4	13	18	14	12	3
-C[T>G]C	24	22	48	27	8	23	31	25	23	5
-C[T>G]G	33	25	59	22	10	24	36	33	28	6
-C[T>G]T	40	44	86	274	13	47	59	39	44	7
-G[C>A]A	253	55	111	29	93	56	73	56	206	11
-G[C>A]C	67	39	82	20	23	40	53	42	59	7
-G[C>A]G	25	11	24	7	9	11	15	13	21	3
-G[C>A]T	148	38	77	18	53	40	51	37	123	6
-G[C>G]A	22	21	47	13	7	21	29	25	20	5
-G[C>G]C	25	23	51	13	8	22	32	28	23	5
-G[C>G]G	8	6	14	4	3	6	8	8	7	2
-G[C>G]T	33	30	67	18	10	29	41	36	29	7
-G[C>T]A	109	80	205	57	41	74	107	109	88	22
-G[C>T]C	111	72	215	56	35	63	106	116	83	21
-G[C>T]G	203	327	273	276	371	330	135	254	258	168
-G[C>T]T	95	68	176	49	30	62	98	100	75	18
-G[T>A]A	19	14	34	10	6	13	20	20	15	4
-G[T>A]C	17	14	30	8	5	14	19	16	15	3
-G[T>A]G	22	19	41	11	7	18	26	22	20	4
-G[T>A]T	21	20	38	9	7	21	26	18	22	3
-G[T>C]A	64	45	119	38	21	38	66	77	46	14
-G[T>C]C	40	30	71	23	15	27	41	44	32	9
-G[T>C]G	74	46	125	41	25	39	68	82	53	15
-G[T>C]T	66	45	118	36	21	40	67	74	50	13
-G[T>G]A	13	11	25	7	4	10	15	14	11	2
-G[T>G]C	8	9	18	7	2	9	12	9	8	2
-G[T>G]G	35	22	51	15	12	20	30	29	29	5
-G[T>G]T	14	17	31	59	5	18	22	13	17	2
-T[C>A]A	201	84	128	38	71	94	93	51	171	13
-T[C>A]C	134	73	123	33	50	79	86	55	121	12
-T[C>A]G	38	13	26	9	13	14	16	14	30	3
-T[C>A]T	299	84	155	42	107	89	105	72	246	14
-T[C>G]A	143	86	110	92	46	88	59	75	68	28
-T[C>G]C	63	56	95	34	21	59	62	48	52	11
-T[C>G]G	14	9	18	8	5	9	10	11	9	3
-T[C>G]T	184	129	177	118	61	135	104	107	103	37
-T[C>T]A	224	111	408	89	38	114	148	213	112	267
-T[C>T]C	144	100	314	67	39	97	140	143	108	67
-T[C>T]G	163	191	231	167	203	187	108	189	164	114
-T[C>T]T	162	94	311	74	31	92	131	170	91	158
-T[T>A]A	52	38	69	24	27	38	43	40	48	11
-T[T>A]C	24	31	56	11	8	34	40	22	30	4
-T[T>A]G	25	21	48	12	8	21	30	26	22	5
-T[T>A]T	53	67	118	26	20	74	85	49	65	10
-T[T>C]A	83	61	157	47	26	54	90	97	63	17
-T[T>C]C	56	42	92	29	23	40	55	55	48	12
-T[T>C]G	62	41	108	34	20	35	61	69	46	12
-T[T>C]T	91	68	164	53	34	61	93	103	71	21
-T[T>G]A	24	30	59	14	8	31	40	28	26	5
-T[T>G]C	33	39	76	21	10	41	52	35	36	6
-T[T>G]G	34	34	73	21	11	35	47	38	33	7
-T[T>G]T	69	83	165	92	22	88	111	73	78	13

SigProfilerAssignment-0.0.30/SigProfilerAssignment/version.py

@@ -1,8 +0,0 @@
-
-# THIS FILE IS GENERATED FROM SigProfilerAssignment SETUP.PY
-short_version = '0.0.30'
-version = '0.0.30'
-Update = 'v0.0.30: Sample reconstruction plot fix and pandas version fix.'
-
-    
-    

SigProfilerAssignment-0.0.30/setup.py

@@ -1,58 +0,0 @@
-from setuptools import setup
-import shutil
-import os
-
-#remove the dist folder first if exists
-if os.path.exists("dist"):
-    shutil.rmtree("dist")
-
-VERSION = '0.0.30'
-
-
-def write_version_py(filename='SigProfilerAssignment/version.py'):
-    # Copied from numpy setup.py
-    cnt = """
-# THIS FILE IS GENERATED FROM SigProfilerAssignment SETUP.PY
-short_version = '%(version)s'
-version = '%(version)s'
-Update = 'v0.0.30: Sample reconstruction plot fix and pandas version fix.'
-
-    
-    """
-    fh = open(filename, 'w')
-    fh.write(cnt % {'version': VERSION,})
-    fh.close()
-
-with open('README.md') as f:
-	long_description = f.read()
-
-requirements=[
-          'scipy>=1.6.3',
-          'numpy>=1.21.2',
-          'pandas>=1.2.4,<2.0.0',
-          'SigProfilerMatrixGenerator>=1.2.14', 
-          'sigProfilerPlotting>=1.3.11',
-          'statsmodels>=0.9.0',
-          'scikit-learn>=0.24.2',
-          'psutil>=5.6.1',
-          'reportlab>=3.5.42',
-          'PyPDF2>=3.0.0',
-          'alive_progress>=2.4.1',
-          'pdf2image>=1.16.0',
-          'PyMuPDF>=1.21.0'
-           ]
-
-write_version_py() 
-setup(name='SigProfilerAssignment',
-      version=VERSION,
-      description='Mutational signatures attribution and decomposition tool',
-      long_description=long_description,
-      long_description_content_type='text/markdown',  # This is important!	
-      url="https://github.com/AlexandrovLab/SigProfilerAssignment.git",
-      author='Raviteja Vangara',
-      author_email='rvangara@health.ucsd.edu',
-      license='UCSD',
-      packages=['SigProfilerAssignment'],
-      install_requires=requirements,
-      include_package_data=True,      
-      zip_safe=False)