bscampp 1.0.6__py3-none-any.whl → 1.0.8__py3-none-any.whl

bscampp/__init__.py

@@ -12,7 +12,7 @@ import logging, os
 # not really needed for BSCAMPP but safe to update here
 os.sys.setrecursionlimit(1000000)
 
-__version__ = "1.0.6"
+__version__ = "1.0.8"
 _INSTALL_PATH = __path__[0]
 
 # global variables to store all loggers

bscampp/configs.py

@@ -4,6 +4,8 @@ try:
 except ImportError:
     from ConfigParser import configparser
 from argparse import ArgumentParser, Namespace
+import subprocess
+
 from bscampp.init_configs import init_config_file
 from bscampp import get_logger, log_exception
 #from bscampp.utils import inferDataType
@@ -22,7 +24,6 @@ _LOG = get_logger(__name__)
 Configuration defined by users and by default values
 '''
 class Configs:
-    global _root_dir
 
     # basic input paths
     info_path = None        # info file for pplacer or EPA-ng
@@ -117,6 +118,24 @@ def _read_config_file(filename, cparser, opts,
             setattr(opts, section, section_name_space)
     return config_defaults
 
+# validate a given binary file is executable
+def validate_binary_executable(name, path):
+    # 1. make sure path exists
+    if not os.path.exists(path):
+        raise FileNotFoundError(f"{path} is not found!")
+
+    # 2. make sure path is executable with no IO error
+    try:
+        p = subprocess.Popen([path], text=True, stdout=subprocess.PIPE,
+                stderr=subprocess.PIPE)
+        _, __ = p.communicate()
+    except OSError:
+        raise AssertionError(' '.join([
+            f"{path} is not executable! Please do the followings:",
+            f"\n\n\t1. obtain a working executable for the software: {name}",
+            f"\n\t2. modify {main_config_path} to replace the old executable path\n",
+            ]))
+
 '''
 Build Config class
 '''
@@ -170,6 +189,8 @@ def buildConfigs(parser, cmdline_args, child_process=False, rerun=False):
 
     # sanity check for existence of base placement binary path
     if Configs.placement_method == 'epa-ng':
-        assert os.path.exists(Configs.epang_path), 'epa-ng not detected!'
+        validate_binary_executable(Configs.placement_method, Configs.epang_path)
+        #assert os.path.exists(Configs.epang_path), 'epa-ng not detected!'
     elif Configs.placement_method == 'pplacer':
-        assert os.path.exists(Configs.pplacer_path), 'pplacer not detected!'
+        validate_binary_executable(Configs.placement_method, Configs.pplacer_path)
+        #assert os.path.exists(Configs.pplacer_path), 'pplacer not detected!'

bscampp/functions.py

@@ -11,6 +11,10 @@ import bscampp.utils as utils
 
 import concurrent.futures
 
+# suppress userwarning when doing subtree suppress_unifurcations
+import warnings
+warnings.filterwarnings("ignore", category=UserWarning)
+
 _LOG = get_logger(__name__)
 
 ############################# helper functions ################################
@@ -29,6 +33,8 @@ def recompileBinariesFromDir(dir):
 
     if cmake_p.returncode != 0:
         _LOG.error("cmake failed!")
+        print("STDOUT:", cmake_stdout)
+        print("STDERR:", cmake_stderr)
         exit(cmake_p.returncode)
     else:
         _LOG.warning("cmake succeeded!")
@@ -59,14 +65,26 @@ def ensureBinaryExecutable(binpath):
         _LOG.warning(f"{binpath} does not exist!")
         b_recompile = True
     else:
-        p = subprocess.Popen([binpath], stdout=subprocess.PIPE,
-                stderr=subprocess.PIPE)
-        stdout, stderr = p.communicate()
+        """
+            added @ 6.13.2025 by Chengze Shen
+            - try-catch OSError to indicate that the binary files
+            - are not executable on the current sytem and need to be
+            - recompiled
+        """
+        try:
+            p = subprocess.Popen([binpath], stdout=subprocess.PIPE,
+                    stderr=subprocess.PIPE)
+            stdout, stderr = p.communicate()
+            returncode = p.returncode
+        except OSError as e:
+            # indicating we need to recompile: anything other than 255 or -1
+            returncode = 7
+
         # 255 or -1 indicates that the binaries work
-        if p.returncode == 255 or p.returncode == -1:
+        if returncode == 255 or returncode == -1:
             pass
         else:
-            _LOG.warning(f"{binpath} return code is {p.returncode}!")
+            _LOG.warning(f"{binpath} return code is {returncode}!")
             b_recompile = True
 
     if b_recompile:
@@ -405,7 +423,8 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
         aln, qaln, cmdline_args, workdir, qname_map, qname_map_rev, 
         pool, lock, dry_run=False):
     t0 = time.perf_counter()
-    _LOG.info('Performing placement on each subtree...')
+    _LOG.info("Performing placement on each subtree with {}...".format(
+        Configs.placement_method))
 
     if dry_run:
         return dict()
@@ -461,7 +480,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
             job = EPAngJob(path=Configs.epang_path,
                     info_path=Configs.info_path, tree_path=tmp_tree,
                     aln_path=tmp_aln, qaln_path=tmp_qaln,
-                    outdir=subtree_dir, num_cpus=Configs.num_cpus)
+                    outdir=subtree_dir, num_cpus=Configs.cpus_per_job)
             jobs.append(job)
             ## for EPA-ng, ensure that outpath name is changed to the one we want
             #_outpath = job.run(logging=f'subtree_{final_subtree_count}')
@@ -481,7 +500,7 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
             job = PplacerTaxtasticJob(path=Configs.pplacer_path,
                     refpkg_dir=refpkg_dir,
                     #molecule=Configs.molecule, model=Configs.model,
-                    outpath=tmp_output, num_cpus=Configs.num_cpus,
+                    outpath=tmp_output, num_cpus=Configs.cpus_per_job,
                     qaln_path=tmp_qaln)
             #tmp_output = job.run(logging=f'subtree_{final_subtree_count}')
             jobs.append(job)
@@ -520,10 +539,22 @@ def placeQueriesToSubtrees(tree, leaf_dict, new_subtree_dict, placed_query_list,
             field_to_idx = {field: i for i, field in enumerate(fields)}
 
             for tmp_place in place_json["placements"]:
-                # convert qname back using qname_map_rev
-                qname = qname_map_rev[tmp_place[tgt][0]]
-                tmp_place[tgt][0] = qname
-                placed_query_list.append(qname)
+                # Fixed @ 7.7.2025 - Chengze Shen
+                #   - pplacer actually can report multiple items
+                #   - in the ["nm"] field.
+                for _idx in range(len(tmp_place[tgt])):
+                    # convert qname back using qname_map_rev
+                    tmp_name = tmp_place[tgt][_idx]
+
+                    # >EPA-ng: tgt=="n" --> qname is string
+                    if isinstance(tmp_name, str):
+                        qname = qname_map_rev[tmp_name]
+                        tmp_place[tgt][_idx] = qname
+                    # >pplacer: tgt=="nm" --> qname is a list of two fields
+                    elif isinstance(tmp_name, list):
+                        qname = qname_map_rev[tmp_name[0]]
+                        tmp_place[tgt][_idx][0] = qname
+                    placed_query_list.append(qname)
 
                 #placed_query_list.append(tmp_place[tgt][0])
                 for i in range(len(tmp_place["p"])):

bscampp/init_configs.py

@@ -71,11 +71,6 @@ def init_config_file(homepath, rerun=False, prioritize_user_software=True):
     cparser.set('basic', 'hamming_distance_dir',
             os.path.join(tools_dir, 'hamming_distance'))
     
-    # macOS TODO: need to recompile the binaries 
-    if 'macos' in platform_name.lower():
-        cparser.set('basic', 'hamming_distance_dir',
-                os.path.join(tools_dir, 'macOS', 'hamming_distance'))
-    
     # prioritize user's software
     if prioritize_user_software:
         print('Detecting existing software from user\'s environment...')

bscampp/pipeline.py

@@ -180,8 +180,6 @@ def clean_temp_files():
         _LOG.info(f'- Removed {temp}')
 
 def parseArguments(dry_run=False, method="BSCAMPP"):
-    global _root_dir, main_config_path
-
     default_outdir = f"{method.lower()}_output"
     default_outname = f"{method.lower()}_result"
 

{bscampp-1.0.6.dist-info → bscampp-1.0.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: bscampp
-Version: 1.0.6
+Version: 1.0.8
 Summary: BSCAMPP and SCAMPP - Scalable Phylogenetic Placement Tools
 Author-email: Eleanor Wedell <ewedell2@illinois.edu>, Chengze Shen <chengze5@illinois.edu>
 License: MIT License
@@ -49,12 +49,13 @@ Requires-Dist: ConfigParser>=5.0.0
 Requires-Dist: numpy>=1.21.6
 Requires-Dist: treeswift>=1.1.45
 Requires-Dist: taxtastic>=0.9.3
+Dynamic: license-file
 
 # BSCAMPP and SCAMPP - Two Scalable Phylogenetic Placement Methods and Frameworks
+[![PyPI - Version](https://img.shields.io/pypi/v/bscampp)](https://pypi.org/project/bscampp/#history)
 [![PyPI - Python Version](https://img.shields.io/pypi/pyversions/bscampp)](https://pypi.org/project/bscampp/)
-[![PyPI - Version](https://img.shields.io/pypi/v/bscampp?color=blue)](https://pypi.org/project/bscampp/#history)
 [![Build Status](https://img.shields.io/github/actions/workflow/status/ewedell/BSCAMPP/python-package.yml?branch=main&label=build)](https://github.com/ewedell/BSCAMPP/)
-[![PyPI - License](https://img.shields.io/pypi/l/bscampp?color=blue)](https://github.com/ewedell/BSCAMPP/blob/main/LICENSE)
+[![PyPI - License](https://img.shields.io/pypi/l/bscampp)](https://github.com/ewedell/BSCAMPP/blob/main/LICENSE)
 [![Changelog](https://img.shields.io/badge/CHANGELOG-grey)](https://github.com/ewedell/BSCAMPP/blob/main/CHANGELOG.md)
 
 **Table of Contents**
@@ -105,8 +106,11 @@ please contact Eleanor Wedell (ewedell2@illinois.edu).
   EPA-ng and/or pplacer are requirements since BSCAMPP and SCAMPP will use them as the base phylogenetic placement methods.
   By default, the software will search for binary executables of `pplacer` and `epa-ng` in the user's environment when running for the first time.
   We also included a compiled version of `pplacer` for the Linux system under `bscampp/tools`.
-* **C++ OpenMP**:
-  We also use OpenMP to speed up the similarity comparison between sequences using C++, which is required to run the pre-compiled binaries.
+* **(macOS) EPA-ng and pplacer**:
+  For macOS systems, you need to have `EPA-ng` and/or `pplacer` installed in your environment.
+* **cmake and C++ OpenMP**:
+  We also use C++ with OpenMP to speed up the similarity comparison between sequences, which is required to run the pre-compiled binaries.
+  In the case of macOS systems, the software will automatically recompile the binaries using `cmake`, which is required to be installed.
 
 ### (1) Install with `pip`
 The easiest way to install BSCAMPP and SCAMPP is to use `pip install`. This will also install all required Python packages.

{bscampp-1.0.6.dist-info → bscampp-1.0.8.dist-info}/RECORD

@@ -1,10 +1,10 @@
-bscampp/__init__.py,sha256=eDIMYifzKrFdtA3Ac7OvPTyIHUO1ZLgVaM0pKFxxEHE,2289
-bscampp/configs.py,sha256=perl6u5hto6J3JV1JMbsTQ6tqr2uGOk-Z9jfzflid0s,6122
+bscampp/__init__.py,sha256=IB6IZQO2RRGF0EOTQ5wL5zzwFXACvuINO9gD3KV4klA,2289
+bscampp/configs.py,sha256=z40mkOwYTeqNjpcEDtyA4G3LskN86sZWc4Vdk46OywA,7014
 bscampp/default.config,sha256=CEfsUHBy--vwJhEcUuJ0btfuGQWb_lKMVWUIP9f5YGw,112
-bscampp/functions.py,sha256=DGHQJLLzXSghDKbha0LW0YPip_45M6MI4t3zdDpzULI,22448
-bscampp/init_configs.py,sha256=EA9sMN5jWj6zj2b-7tN19LhX2Ef61ByQLxQRLHAqLDM,3600
+bscampp/functions.py,sha256=f5YY9LaIW3LieFvlWvcUlWysEuqVuwJXTmOq5laFxJE,23782
+bscampp/init_configs.py,sha256=hDCH2oatgli2wSmi-uDM96TSiBvLKTAhobBSSKfjl2g,3381
 bscampp/jobs.py,sha256=v7buZJs1AnNoXiILwu-W8fo3QjxAh3i9Mp7xfmlJvAY,7569
-bscampp/pipeline.py,sha256=IPZnXZmVxGGfbVUuGCQh5X9oBq48-6pA9QkuvMGPTag,14000
+bscampp/pipeline.py,sha256=J-RQH54R27m6fhzIpGX0MJuE3ZFk5rcnsROpwC_n5CE,13960
 bscampp/utils.py,sha256=-wns6FaWMKD2wVqjxdBQvjTdagTjywBIaGfqb2mupe4,30039
 bscampp/tools/epa-ng,sha256=f3EVoZAAOXLN6l521qp-TrWDl5J2nqL3tGgjPaQE9WQ,3772096
 bscampp/tools/pplacer,sha256=p0H4eo9uuiYoWS_kJbPfauOV99i7BXJdZSiwXIuLxTw,7834576
@@ -17,9 +17,9 @@ bscampp/tools/hamming_distance/src/fragment_tree_hamming.cpp,sha256=xCmyAT-OZJOD
 bscampp/tools/hamming_distance/src/fragment_tree_hamming_new.cpp,sha256=eKxgODRlpf0hU84QjNhigvRhWCT9tiJZjA5oQFQ1bUk,7404
 bscampp/tools/hamming_distance/src/homology.cpp,sha256=ZE0uXZWQ-cN4U1Wk5kUr_KKHgzsgA6Sno-IViRa4tmI,6053
 bscampp/tools/hamming_distance/src/new_hamming.cpp,sha256=fBRm99RquBZgZjaLOn9xDI3cH9NchhrxKbL-11j8fmk,5342
-bscampp-1.0.6.dist-info/LICENSE,sha256=HEa4YQdOR0e2Gz-NiOwr9X6aJcZtY0AGmlJQDmfN0Iw,1064
-bscampp-1.0.6.dist-info/METADATA,sha256=0sWAKK30wlps8i0d1BdFqyv5MZVgefRnTn_-yMmO8lQ,12602
-bscampp-1.0.6.dist-info/WHEEL,sha256=52BFRY2Up02UkjOa29eZOS2VxUrpPORXg1pkohGGUS8,91
-bscampp-1.0.6.dist-info/entry_points.txt,sha256=4Ft83qHc39tNNpMLgSgFXDHM-vuAB99JtmczCQj5pq8,204
-bscampp-1.0.6.dist-info/top_level.txt,sha256=1loGRUAft6Tcdq0f3lHbVwWN7W_SW1srfhAVSpg9DWE,8
-bscampp-1.0.6.dist-info/RECORD,,
+bscampp-1.0.8.dist-info/licenses/LICENSE,sha256=HEa4YQdOR0e2Gz-NiOwr9X6aJcZtY0AGmlJQDmfN0Iw,1064
+bscampp-1.0.8.dist-info/METADATA,sha256=2w_otUXBGIuvf6nXV-7hY958cxllhTZd6T18rB5nR8Q,12877
+bscampp-1.0.8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+bscampp-1.0.8.dist-info/entry_points.txt,sha256=4Ft83qHc39tNNpMLgSgFXDHM-vuAB99JtmczCQj5pq8,204
+bscampp-1.0.8.dist-info/top_level.txt,sha256=1loGRUAft6Tcdq0f3lHbVwWN7W_SW1srfhAVSpg9DWE,8
+bscampp-1.0.8.dist-info/RECORD,,

{bscampp-1.0.6.dist-info → bscampp-1.0.8.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (76.0.0)
+Generator: setuptools (80.9.0)
 Root-Is-Purelib: true
 Tag: py3-none-any