Merge pull request #195 from SigProfilerSuite/python-312-fix-p2

Python 312 fix p2

Author: mdbarnesUCSD

CHANGELOG.md

@@ -6,6 +6,13 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 
 ## [Unreleased]
 
+## [1.1.3] - 2026-01-23
+### Fixed
+- Fixed Python 3.12 compatibility issues with pandas StringArray in `set_index()` calls:
+  - Replaced `set_index()` with direct index assignment using `pd.Index()` to avoid StringArray unhashable type errors
+  - Fixed index restoration from `devopts` dictionary in both `denovo_refit_option` and `decompose_fit_option` paths
+  - Fixed index assignment in `make_final_solution()` for process matrices, exposure matrices, and signature statistics
+
 ## [1.1.2] - 2026-01-23
 ### Fixed
 - Fixed Python 3.12 compatibility issues with pandas DataFrame/Series indexing and method calls:

SigProfilerAssignment/decompose_subroutines.py

@@ -1125,7 +1125,19 @@ def make_final_solution(
         pass
 
     processAvg = pd.DataFrame(processAvg.astype(float))
-    processes = processAvg.set_index(index)
+    # Convert index to pandas Index to handle StringArray compatibility in Python 3.12+
+    # StringArray is unhashable and cannot be used directly with set_index
+    if isinstance(index, str):
+        processes = processAvg.set_index(index)
+    else:
+        # Convert to list first, then create pandas Index and assign directly
+        if hasattr(index, 'tolist'):
+            index_list = index.tolist()
+        else:
+            index_list = list(index)
+        # Assign index directly instead of using set_index to avoid column lookup
+        processes = processAvg.copy()
+        processes.index = pd.Index(index_list)
     processes.columns = allsigids
     processes = processes.rename_axis("MutationType", axis="columns")
     processes.to_csv(
@@ -1140,8 +1152,17 @@ def make_final_solution(
         index_label=[processes.columns.name],
     )
     exposureAvg = pd.DataFrame(exposureAvg.astype(int))
-    allsigids = np.array(allsigids)
-    exposures = exposureAvg.set_index(allsigids)
+    # Convert allsigids to list to handle StringArray compatibility in Python 3.12+
+    if isinstance(allsigids, str):
+        allsigids_list = allsigids
+    elif hasattr(allsigids, 'tolist'):
+        allsigids_list = allsigids.tolist()
+    else:
+        allsigids_list = list(allsigids)
+    allsigids = np.array(allsigids_list)
+    # Assign index directly instead of using set_index to avoid column lookup
+    exposures = exposureAvg.copy()
+    exposures.index = pd.Index(allsigids_list)
     exposures.columns = allcolnames
     exposures = exposures.T
     exposures = exposures.rename_axis("Samples", axis="columns")
@@ -1258,7 +1279,18 @@ def make_final_solution(
     if refit_denovo_signatures:
         try:
             process_std_error = pd.DataFrame(process_std_error)
-            processSTE = process_std_error.set_index(index)
+            # Convert index to pandas Index to handle StringArray compatibility in Python 3.12+
+            if isinstance(index, str):
+                processSTE = process_std_error.set_index(index)
+            else:
+                # Convert to list first, then create pandas Index and assign directly
+                if hasattr(index, 'tolist'):
+                    index_list = index.tolist()
+                else:
+                    index_list = list(index)
+                # Assign index directly instead of using set_index to avoid column lookup
+                processSTE = process_std_error.copy()
+                processSTE.index = pd.Index(index_list)
             processSTE.columns = allsigids
             processSTE = processSTE.rename_axis("MutationType", axis="columns")
             processSTE.to_csv(
@@ -1276,7 +1308,16 @@ def make_final_solution(
             pass
     if refit_denovo_signatures:
         try:
-            signature_stats = signature_stats.set_index(allsigids)
+            # Convert allsigids to list to handle StringArray compatibility in Python 3.12+
+            if isinstance(allsigids, str):
+                allsigids_list = allsigids
+            elif hasattr(allsigids, 'tolist'):
+                allsigids_list = allsigids.tolist()
+            else:
+                allsigids_list = list(allsigids)
+            # Assign index directly instead of using set_index to avoid column lookup
+            signature_stats = signature_stats.copy()
+            signature_stats.index = pd.Index(allsigids_list)
             signature_stats = signature_stats.rename_axis("Signatures", axis="columns")
             signature_stats.to_csv(
                 layer_directory

SigProfilerAssignment/decomposition.py

@@ -577,7 +577,19 @@ def spa_analyze(
             listOfSignatures = devopts["listOfSignatures"]
             index = devopts["index"]
             colnames = devopts["colnames"]
-            genomes = genomes.set_index(index)
+            # Convert index to pandas Index to handle StringArray compatibility in Python 3.12+
+            # StringArray is unhashable and cannot be used directly with set_index
+            if isinstance(index, str):
+                # If it's a column name, use set_index normally
+                genomes = genomes.set_index(index)
+            else:
+                # Convert to list first, then create pandas Index and assign directly
+                if hasattr(index, 'tolist'):
+                    index_list = index.tolist()
+                else:
+                    index_list = list(index)
+                # Assign index directly instead of using set_index to avoid column lookup
+                genomes.index = pd.Index(index_list)
             genomes.columns = colnames
             # genomes = genomes.rename_axis("Mutation Types", axis="columns")
 
@@ -766,7 +778,19 @@ def spa_analyze(
             listOfSignatures = devopts["listOfSignatures"]
             index = devopts["index"]
             colnames = devopts["colnames"]
-            genomes = genomes.set_index(index)
+            # Convert index to pandas Index to handle StringArray compatibility in Python 3.12+
+            # StringArray is unhashable and cannot be used directly with set_index
+            if isinstance(index, str):
+                # If it's a column name, use set_index normally
+                genomes = genomes.set_index(index)
+            else:
+                # Convert to list first, then create pandas Index and assign directly
+                if hasattr(index, 'tolist'):
+                    index_list = index.tolist()
+                else:
+                    index_list = list(index)
+                # Assign index directly instead of using set_index to avoid column lookup
+                genomes.index = pd.Index(index_list)
             genomes.columns = colnames
             make_decomposition_plots = devopts["make_decomposition_plots"]
             # genomes = genomes.rename_axis("Mutation Types", axis="columns")

setup.py

@@ -6,7 +6,7 @@
 if os.path.exists("dist"):
     shutil.rmtree("dist")
 
-VERSION = "1.1.2"
+VERSION = "1.1.3"
 
 
 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 = 'v1.1.2: Fixed to_csv() calls for Python 3.12 compatibility'
+Update = 'v1.1.3: Fixed set_index() calls for Python 3.12 StringArray compatibility'
 
     """
     fh = open(filename, "w")