Merge pull request #177 from AlexandrovLab/custom-sigs

- Add decomposition plot support for custom signatures

Author: mdbarnesUCSD

CHANGELOG.md

@@ -6,6 +6,17 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 
 ## [Unreleased]
 
+## [0.2.5] - 2025-08-07
+
+### Added
+- Support for generating decomposition plots for custom signature sets.
+
+### Changed
+- Refactored the test script to be more readable and maintainable.
+
+### Fixed
+- Removed redundant plotting of COSMIC signatures in the decomposition plots.
+
 ## [0.2.4] - 2025-08-07
 
 ### Added

SigProfilerAssignment/DecompositionPlots/PlotDecomposition.py

@@ -897,6 +897,7 @@ def run_PlotDecomposition(
     exome=False,
     custom_text=None,
     volume=None,
+    use_custom_basis=False,
 ):
     """
     Generates a decomposition plot of the denovo_mtx using the basis_mtx.
@@ -936,16 +937,48 @@ def run_PlotDecomposition(
     None.
     """
 
-    # Create the denovo plots and load basis plots
-    denovo_plots_dict, basis_plots_dict = gen_sub_plots(
-        denovo_mtx,
-        basis_mtx,
-        output_path,
-        project,
-        mtype,
-        ss_decomp=False,
-        volume=volume,
-    )
+    # --- Efficiently Generate or Load Plots ---
+    # This logic avoids redundant plot generation for COSMIC signatures.
+    if use_custom_basis:
+        # For custom sets, generate plots for BOTH the de novo and basis matrices.
+        denovo_plots_dict, basis_plots_dict = gen_sub_plots(
+            denovo_mtx,
+            basis_mtx,
+            output_path,
+            project,
+            mtype,
+            ss_decomp=False,
+            volume=volume,
+        )
+    else:  # This is the standard COSMIC case
+        # For COSMIC sets, ONLY generate plots for the de novo matrix to save time.
+        # The basis plots will be loaded from the pre-computed cache.
+        # We assume gen_sub_plots can handle a None value for the second matrix
+        # and we use a placeholder for the second return value.
+        denovo_plots_dict, _ = gen_sub_plots(
+            denovo_mtx,
+            None,  # Pass None to avoid generating plots for the basis_mtx
+            output_path,
+            project,
+            mtype,
+            ss_decomp=False,
+            volume=volume,
+        )
+
+        # Now, load the real basis plots from the cache.
+        basis_plots_dict = install_cosmic_plots(
+            context_type=mtype,
+            genome_build=genome_build,
+            cosmic_version=cosmic_version,
+            exome=exome,
+            volume=volume,
+        )
+        # And filter for only the specific signatures present in this decomposition.
+        basis_plots_dict = {key: basis_plots_dict[key] for key in basis_names}
+
+    # --- Continue with Plot Assembly ---
+
+    # Generate the reconstructed profile plot
     reconstructed_mtx, reconstruction_plot_dict = gen_reconstructed_png_percent(
         denovo_name,
         basis_mtx,
@@ -956,26 +989,20 @@ def run_PlotDecomposition(
         mtype,
         volume=volume,
     )
-    # Create a subset matrix with the present signatures
+
+    # Create a subset matrix with the present signatures for statistics
     present_sigs = np.array(basis_mtx[basis_names])
     reconstructed_mtx = np.dot(present_sigs, nonzero_exposures)
-    # Convert dictionary of bytes to dictionary of images
+
+    # Convert all plot dictionaries from bytes to image objects for assembly
     denovo_plots_dict = convert_to_imgReaderDict(denovo_plots_dict)
-    # Load in the COSMIC plots
-    basis_plots_dict = install_cosmic_plots(
-        context_type=mtype,
-        genome_build=genome_build,
-        cosmic_version=cosmic_version,
-        exome=exome,
-        volume=volume,
-    )
-    basis_plots_dict = {key: basis_plots_dict[key] for key in basis_names}
     basis_plots_dict = convert_to_imgReaderDict(basis_plots_dict)
-    # Generate the reconstruction plot
     reconstruction_plot_dict = convert_to_imgReaderDict(reconstruction_plot_dict)
+
     # Get the reconstruction statistics
     statistics = calculate_similarities(denovo_mtx, denovo_name, reconstructed_mtx)
-    # Return the decomposition plot as a byte array
+
+    # Generate the final decomposition plot as a byte array
     byte_plot = gen_decomposition(
         denovo_name,
         basis_names,
@@ -991,6 +1018,7 @@ def run_PlotDecomposition(
         cosmic_version=cosmic_version,
         custom_text=custom_text,
     )
+
     # Clear the plotting memory
     sigPlt.clear_plotting_memory()
 

SigProfilerAssignment/data/tests/txt_input/custom-SBS-list.txt

@@ -0,0 +1,97 @@
+Type	SpecialSignatureOne	SpecialSignatureA	SpecialSignatureAB	SpecialSignatureABC	SpecialSignatureABCD	SpecialSignatureABCDE	SpecialSignatureABCDEF	SpecialSignatureABCDEFG	SpecialSignatureABCDEFGH
+A[C>A]A	0.000886157	5.80E-07	0.020808323	0.042196498	0.0119976	0.000425233	6.70E-05	0.002329386	0.004830406
+A[C>A]C	0.002280405	0.000148004	0.016506603	0.033297236	0.009438112	0.000524287	0.000179116	0.000460879	0.001150097
+A[C>A]G	0.000177031	5.23E-05	0.0017507	0.015598705	0.00184963	5.20E-05	7.12E-05	0.000185951	0.000377032
+A[C>A]T	0.001280227	9.78E-05	0.012204882	0.029497552	0.006608678	0.000180099	0.000248161	0.000709813	0.001960165
+A[C>G]A	0.00186033	2.23E-16	0.019707883	0.006889428	0.01009798	0.000471258	6.49E-05	8.56E-06	0.001120094
+A[C>G]C	0.001220217	0.000133004	0.011704682	0.002839764	0.00569886	0.000289158	2.16E-05	0.00018895	0.000319027
+A[C>G]G	0.00011502	1.52E-05	0.000253101	0.001279894	0.001719656	2.22E-16	5.08E-05	0.000504867	2.22E-16
+A[C>G]T	0.001140202	9.12E-05	0.017406963	0.003549705	0.01009798	0.000254139	9.57E-05	0.000284925	0.001930162
+A[C>T]A	0.025004441	6.11E-05	0.014205682	0.008699278	0.032593481	0.06413516	0.000122079	0.001409629	0.001330112
+A[C>T]C	0.006321123	0.00138004	0.012404962	0.004179653	0.017896421	0.022912561	0.017711498	0.038089959	0.007680645
+A[C>T]G	0.36506483	3.27E-05	0.002571028	0.000782935	0.006178764	0.120065821	2.23E-16	0.000542857	0.000927078
+A[C>T]T	0.009581702	0.001860054	0.012104842	0.004249647	0.021995601	0.021311683	0.007424819	0.010797154	0.007480629
+A[T>A]A	0.000800142	9.59E-05	0.005492197	0.009419218	0.007378524	9.83E-05	0.00194126	0.002489344	0.012401042
+A[T>A]C	0.002230396	0.000878025	0.007212885	0.003529707	0.006928614	0.000546299	0.000670435	0.000466877	0.003420287
+A[T>A]G	0.001140202	2.55E-05	0.009643858	0.012398971	0.009118176	2.22E-16	0.000587381	0.001559589	0.000211018
+A[T>A]T	0.000183032	2.23E-16	0.006112445	0.004209651	0.00619876	0.00166091	0.00371241	0.008287815	0.050704262
+A[T>C]A	0.001090193	6.17E-05	0.016506603	0.00794934	0.046190762	0.002211213	0.000490318	0.001019731	0.000288024
+A[T>C]C	0.003040539	2.15E-05	0.007763105	0.000714941	0.013397321	0.00113062	0.000227147	0.00045588	0.000533045
+A[T>C]G	0.000106019	1.32E-05	0.012304922	0.004999585	0.038192362	0.004242326	0.000187121	0.001499605	0.00107009
+A[T>C]T	0.00574102	0.000155004	0.017306923	0.001559871	0.038292342	0.000502275	0.000615399	0.002149434	0.002650223
+A[T>G]A	0.000172031	0.000238007	0.003951581	0.001049913	0.003859228	2.57E-05	0.000473307	8.44E-05	0.011901
+A[T>G]C	0.000207037	7.46E-05	0.00260104	0.000158987	0.00259948	0.000204112	0.00024716	0.000931754	0.001050088
+A[T>G]G	0.000268048	2.05E-06	0.006302521	0.001419882	0.007938412	0.000107059	0.000276179	0.001469613	0.000604051
+A[T>G]T	0.00011202	3.76E-06	0.003971589	0.000215982	0.004829034	0.000386212	0.000309201	0.000732807	2.22E-16
+C[C>A]A	0.000312055	0.000208006	0.022509004	0.080693302	0.007428514	0.001820998	0.000455295	0.0011397	0.000109009
+C[C>A]C	0.001790318	9.53E-05	0.025310124	0.079693385	0.006138772	0.004092243	0.000240156	0.001549592	0.00309026
+C[C>A]G	9.32E-05	2.23E-16	0.002511004	0.024497967	0.003459308	0.001650905	2.23E-16	0.000407893	0.001910161
+C[C>A]T	2.23E-16	0.000421012	0.015406162	0.068994273	0.006488702	0.009605265	0.000372242	0.001489607	0.013501135
+C[C>G]A	2.41E-05	2.23E-16	0.019307723	0.00782935	0.006958608	4.11E-05	2.23E-16	0.000457879	0.000206017
+C[C>G]C	7.68E-05	0.00035101	0.014105642	0.00686943	0.009068186	3.72E-05	0.000131085	0.000959747	0.00226019
+C[C>G]G	0.000352062	4.87E-05	0.0015006	0.003399718	0.002489502	8.43E-05	7.76E-05	0.000790792	0.000262022
+C[C>G]T	2.23E-16	6.36E-05	0.022308924	0.006819434	0.009528094	1.44E-05	9.57E-05	0.000889766	0.004240356
+C[C>T]A	0.002000355	0.004320125	0.016106443	0.012698946	0.019396121	0.01360746	0.050032478	0.106971801	0.002710228
+C[C>T]C	0.000270048	0.000170005	0.020208083	0.011499046	0.018996201	0.014808118	0.05533592	0.181952034	0.005970502
+C[C>T]G	0.196034813	0.00277008	0.002240896	0.003309725	0.017496501	0.122066918	0.006714358	0.022094175	0.008940751
+C[C>T]T	0.000196035	0.001450042	0.023009204	0.015998672	0.023095381	0.01380757	0.056636764	0.125966794	0.020601731
+C[T>A]A	4.30E-05	0.000243007	0.007943177	0.014398805	0.003539292	6.66E-05	0.000177115	0.001119705	0.003470292
+C[T>A]C	0.00039307	0.000244007	0.017807123	0.012198987	0.005188962	0.000551302	0.000261169	0.000957748	0.013401126
+C[T>A]G	0.000324057	0.000149004	0.013205282	0.032497303	0.005008998	0.000115063	0.000112073	0.000882767	0.000961081
+C[T>A]T	0.000260046	2.23E-16	0.012705082	0.011799021	0.00429914	0.000179098	0.000371241	0.00011197	0.048704094
+C[T>C]A	2.23E-16	0.000302009	0.008833533	0.00517957	0.013297341	0.00166091	0.000451293	0.001529597	0.007570636
+C[T>C]C	0.002500444	2.85E-05	0.014505802	0.002359804	0.01039792	0.001030565	0.000259168	0.000839779	0.003580301
+C[T>C]G	0.000360064	2.23E-16	0.010104042	0.005499544	0.020795841	0.005322918	0.000104068	0.000729808	0.002170182
+C[T>C]T	4.26E-05	0.000186005	0.015806323	0.002559788	0.014897021	0.001270696	0.001821182	0.00572849	0.043103622
+C[T>G]A	3.55E-05	0.000153004	0.004371749	0.000869928	0.002589482	4.95E-05	8.43E-05	0.000358905	0.003280276
+C[T>G]C	0.000212038	0.000155004	0.007042817	0.000440963	0.004609078	0.00078443	0.000172112	0.000791791	0.005580469
+C[T>G]G	0.000128023	0.000121003	0.010704282	0.003789685	0.006128774	0.000924507	2.23E-16	0.00128966	0.0011901
+C[T>G]T	0.00017103	0.000203006	0.007032813	0.0011999	0.00729854	0.001300713	2.65E-05	3.68E-06	0.003400286
+G[C>A]A	0.00158028	8.62E-05	0.007132853	0.031597377	0.01019796	0.000427234	4.06E-05	0.000745803	0.001050088
+G[C>A]C	0.00033906	2.23E-16	0.010904362	0.034797112	0.00764847	0.000913501	0.00013809	0.000214943	0.000742062
+G[C>A]G	0.000587104	1.39E-05	0.001430572	0.01529873	0.002339532	0.000357196	4.25E-05	7.29E-05	2.22E-16
+G[C>A]T	2.23E-16	5.12E-05	0.010004002	0.021198241	0.006818636	0.001230674	2.23E-16	0.000146961	0.000363031
+G[C>G]A	9.59E-06	6.54E-05	0.012705082	0.005289561	0.004679064	1.94E-05	1.96E-05	5.11E-05	0.000182015
+G[C>G]C	0.000164029	0.000261008	0.008633453	0.005749523	0.005078984	0.000685376	3.32E-05	0.000451881	0.00047304
+G[C>G]G	0.000166029	2.23E-16	0.002420968	0.003299726	0.001509698	2.22E-16	4.51E-05	0.000336911	0.000180015
+G[C>G]T	2.23E-16	0.000134004	0.014705882	0.003039748	0.006718656	0.000321176	4.20E-05	0.000276927	0.001280108
+G[C>T]A	0.004440789	2.23E-16	0.016406563	0.004089661	0.019396121	0.09365134	8.50E-05	0.000854775	0.000157013
+G[C>T]C	9.28E-05	1.29E-07	0.013405362	0.00445963	0.021595681	0.111060885	0.011707599	0.026293069	9.51E-06
+G[C>T]G	0.218038721	2.23E-16	0.000528211	0.00048196	0.01229754	0.178097635	5.52E-06	0.000175954	3.43E-05
+G[C>T]T	3.84E-05	5.04E-05	0.011304522	0.003809684	0.018596281	0.074440809	0.016110458	0.00538858	0.000758064
+G[T>A]A	8.12E-05	2.23E-16	0.007062825	0.009219235	0.003669266	5.10E-05	0.00040126	0.000525861	0.001730145
+G[T>A]C	0.000129023	0.000131004	0.007482993	0.003599701	0.002909418	0.000489268	0.000224145	0.000255933	0.002270191
+G[T>A]G	0.000246044	6.38E-05	0.011604642	0.013098913	0.004109178	0.000195107	0.000290188	0.000655827	2.22E-16
+G[T>A]T	0.000258046	6.28E-05	0.012304922	0.005519542	0.003219356	0.000248136	0.000306199	5.59E-05	0.020301706
+G[T>C]A	0.001050186	2.23E-16	0.009983994	0.002899759	0.014097181	0.00257141	0.000101066	0.00026593	2.22E-16
+G[T>C]C	0.001900337	2.74E-05	0.005642257	0.001179902	0.007938412	0.001110609	7.41E-05	2.22E-16	0.000423036
+G[T>C]G	0.001170208	0.000107003	0.010904362	0.002369803	0.014997001	0.002451344	7.25E-05	0.000528861	0.000232019
+G[T>C]T	7.13E-05	1.25E-05	0.010204082	0.000754937	0.013697261	0.000906497	0.000811527	0.006848194	0.007070594
+G[T>G]A	2.23E-16	0.000209006	0.004421769	0.00059795	0.002519496	4.35E-05	9.07E-05	0.000322915	0.002480208
+G[T>G]C	2.23E-16	0.000133004	0.002340936	0.000252979	0.001709658	0.000213117	0.000121079	0.000320915	0.00166014
+G[T>G]G	0.000348062	3.71E-05	0.010904362	0.002429798	0.00529894	0.00027415	0.000325211	0.002409365	0.006030507
+G[T>G]T	1.46E-05	2.23E-16	0.005832333	0.000251979	0.00234953	0.000787431	0.000835542	0.001829518	0.006750567
+T[C>A]A	6.58E-05	0.000639018	0.008413365	0.039796697	0.007858428	0.000356195	0.000390253	0.002899236	0.002520212
+T[C>A]C	0.002530449	0.000170005	0.016806723	0.046896108	0.009168166	0.000498273	0.000636413	0.003519073	0.014201193
+T[C>A]G	2.23E-16	8.53E-05	0.001410564	0.010899095	0.00229954	0.00020011	3.44E-05	0.000299921	2.22E-16
+T[C>A]T	5.88E-06	0.000440013	0.00997399	0.041296572	0.012697461	0.00074741	0.000613398	0.001409629	0.003730313
+T[C>G]A	2.23E-16	2.23E-16	0.013305322	0.003579703	0.007808438	4.97E-05	2.23E-16	0.000253933	0.001940163
+T[C>G]C	0.000203036	9.87E-05	0.020308123	0.005949506	0.00759848	0.000113062	5.09E-05	0.001619573	0.004610387
+T[C>G]G	6.94E-05	4.12E-05	0.000520208	0.001829848	0.001709658	1.95E-06	6.35E-06	0.000363904	8.75E-06
+T[C>G]T	0.001560277	0.001180034	0.023509404	0.003379719	0.0124975	0.00012807	0.000281182	0.000713812	0.012301034
+T[C>T]A	0.001110197	0.53601551	0.005762305	0.00373969	0.020295941	0.012006582	0.238154593	0.006608258	0.017301454
+T[C>T]C	3.73E-05	0.097302815	0.013305322	0.00698942	0.022995401	0.01050576	0.331215001	0.210944392	0.013701151
+T[C>T]G	0.110019538	0.044201279	0.000989396	0.000280977	0.015896821	0.052028522	0.074148132	0.057384873	4.64E-05
+T[C>T]T	2.23E-16	0.300008682	0.008173269	0.007219401	0.020995801	0.006993834	0.107069502	0.101973118	0.002320195
+T[T>A]A	0.006721194	0.000422012	0.006512605	0.008879263	0.006308738	0.000454249	0.000356231	0.003239146	0.015401294
+T[T>A]C	2.23E-16	0.000192006	0.010504202	0.004309642	0.004139172	0.000114063	0.000389253	0.00223941	0.034202874
+T[T>A]G	2.80E-05	5.01E-05	0.006712685	0.009539208	0.004789042	2.95E-05	2.23E-16	0.001699552	0.012201025
+T[T>A]T	0.002250399	0.000205006	0.013005202	0.005619534	0.00874825	0.000426234	0.000798518	2.22E-16	0.279023452
+T[T>C]A	0.000255045	2.23E-16	0.015206082	0.0024098	0.017996401	0.001480811	0.002151396	0.005998419	0.030202538
+T[T>C]C	0.003390603	4.35E-05	0.008993597	0.000502958	0.009708058	0.001090598	0.000596387	0.001729544	0.020901757
+T[T>C]G	0.000416074	0.000117003	0.006932773	0.001739856	0.012697461	0.002131169	0.000319207	0.003808996	0.012901084
+T[T>C]T	0.004330769	3.58E-05	0.013905562	0.000635947	0.018496301	0.001740954	0.000812527	0.002969217	0.088107406
+T[T>G]A	2.23E-16	1.67E-05	0.007252901	0.000376969	0.005218956	0.000105058	0.000128083	0.000954748	0.019301622
+T[T>G]C	5.51E-05	7.04E-05	0.006282513	0.000173986	0.006558688	0.000287157	0.000116075	0.001549592	0.017401462
+T[T>G]G	0.000583103	9.54E-05	0.008053221	0.002319807	0.006938612	0.000324178	2.23E-16	0.001349644	0.007640642
+T[T>G]T	2.23E-16	2.23E-16	0.010504202	0.000567953	0.013497301	0.001010554	8.30E-05	0.001769534	0.021701824

SigProfilerAssignment/decompose_subroutines.py

@@ -574,7 +574,6 @@ def signature_decomposition(
         if (
             mtype_par != "none"
             and make_decomposition_plots == True
-            and signature_database is None
         ):
             # reformat the first column of cosmic signature dataframe
             cosmic_sigs_DF = sigDatabases_DF.copy(deep=True)
@@ -605,6 +604,7 @@ def signature_decomposition(
                 genome_build=genome_build,
                 exome=exome,
                 volume=volume,
+                use_custom_basis=(signature_database is not None)
             )
 
             byte_plot.seek(0)

setup.py

@@ -6,7 +6,7 @@
 if os.path.exists("dist"):
     shutil.rmtree("dist")
 
-VERSION = "0.2.4"
+VERSION = "0.2.5"
 
 
 def write_version_py(filename="SigProfilerAssignment/version.py"):
@@ -15,7 +15,7 @@ def write_version_py(filename="SigProfilerAssignment/version.py"):
 # THIS FILE IS GENERATED FROM SigProfilerAssignment SETUP.PY
 short_version = '%(version)s'
 version = '%(version)s'
-Update = 'v0.2.4:  Add rn7 and mm39 reference signatures'
+Update = 'v0.2.5:  Support for generating decomposition plots for custom signature sets.'
 
     """
     fh = open(filename, "w")