libgunshotmatch 0.9.0__py3-none-any.whl → 0.10.0__py3-none-any.whl

libgunshotmatch/__init__.py

@@ -29,5 +29,5 @@ Base library for GunShotMatch.
 __author__: str = "Dominic Davis-Foster"
 __copyright__: str = "2020-2023 Dominic Davis-Foster"
 __license__: str = "MIT License"
-__version__: str = "0.9.0"
+__version__: str = "0.10.0"
 __email__: str = "dominic@davis-foster.co.uk"

libgunshotmatch/consolidate/__init__.py

@@ -59,7 +59,8 @@ __all__ = (
 		"ConsolidatedSearchResult",
 		"match_counter",
 		"pairwise_ms_comparisons",
-		"ConsolidatedPeakFilter"
+		"ConsolidatedPeakFilter",
+		"InvertedFilter",
 		)
 
 
@@ -205,12 +206,16 @@ class ConsolidatedSearchResult:
 		:param d:
 		"""
 
+		hit_numbers = [float("nan") if hn == -65535 else hn for hn in d["hit_numbers"]]
+		mf_list = [float("nan") if mf == -65535 else mf for mf in d["mf_list"]]
+		rmf_list = [float("nan") if mf == -65535 else mf for mf in d["rmf_list"]]
+
 		return cls(
 				name=d["name"],
 				cas=d["cas"],
-				mf_list=d["mf_list"],
-				rmf_list=d["rmf_list"],
-				hit_numbers=d["hit_numbers"],
+				mf_list=mf_list,
+				rmf_list=rmf_list,
+				hit_numbers=hit_numbers,
 				reference_data=d["reference_data"],
 				)
 
@@ -768,3 +773,34 @@ class ConsolidatedPeakFilter:
 		"""
 
 		return [cp for cp in consolidated_peaks if not self.should_filter_peak(cp)]
+
+
+@_fix_init_annotations
+@attr.define
+class InvertedFilter(ConsolidatedPeakFilter):
+	"""
+	Inverted version of :class:`~.ConsolidatedPeakFilter`.
+
+	Returns peaks which would be excluded by a :class:`~.ConsolidatedPeakFilter`.
+
+	.. versionadded:: 0.10.0
+	"""
+
+	def print_skip_reason(self, peak: ConsolidatedPeak, reason: str) -> None:  # noqa: D102
+		if self.verbose:
+			print(f"Would reject peak at {peak.rt / 60:0.3f} mins:", reason)
+
+	def filter(self, consolidated_peaks: List[ConsolidatedPeak]) -> List[ConsolidatedPeak]:  # noqa: A003  # pylint: disable=redefined-builtin
+		"""
+		Filter a list of consolidated peaks.
+
+		:param consolidated_peaks:
+		"""
+
+		filtered_consolidated_peaks = []
+
+		for cp in consolidated_peaks:
+			if self.should_filter_peak(cp):
+				filtered_consolidated_peaks.append(cp)
+
+		return filtered_consolidated_peaks

libgunshotmatch/gzip_util.py

@@ -51,7 +51,13 @@ def read_gzip_json(path: PathLike) -> JSONOutput:
 	"""
 
 	with gzip.open(PathPlus(path), 'r') as f:
-		return sdjson.load(f)
+		try:
+			# 3rd party
+			import orjson
+			data = f.read().decode().replace("NaN", "-65535")
+			return orjson.loads(data)
+		except ImportError:
+			return sdjson.load(f)
 
 
 def write_gzip_json(path: PathLike, data: JSONInput, indent: Optional[int] = 2) -> None:

libgunshotmatch/project.py

@@ -50,7 +50,7 @@ from libgunshotmatch.datafile import Repeat
 from libgunshotmatch.peak import PeakList, QualifiedPeak, peak_from_dict
 from libgunshotmatch.utils import create_alignment
 
-__all__ = ["Project"]
+__all__ = ("Project", "consolidate")
 
 
 @attr.define
@@ -176,31 +176,7 @@ class Project:
 			between the repeats for each aligned peak.
 		"""
 
-		ms_comparison_df = pairwise_ms_comparisons(self.alignment)
-
-		peak_numbers: List[int] = []
-		peak: Optional[QualifiedPeak]
-
-		qualified_peak_array = []
-
-		# for experiment in project.alignment.expr_code:
-		for experiment in self.datafile_data:
-			qualified_peaks = self.datafile_data[experiment].qualified_peaks
-			assert qualified_peaks is not None
-			for peak in qualified_peaks:
-				assert peak.peak_number is not None
-				peak_numbers.append(peak.peak_number)
-			qualified_peak_array.append(qualified_peaks)
-
-		# Convert peak_numbers to a set and sort smallest to largest
-		peak_numbers = sorted(set(peak_numbers))
-
-		consolidated_peaks = match_counter(
-				engine=engine,
-				peak_numbers=peak_numbers,
-				qualified_peaks=qualified_peak_array,
-				ms_comp_data=ms_comparison_df,
-				)
+		consolidated_peaks, ms_comparison_df = consolidate(self, engine)
 
 		if peak_filter is None:
 			self.consolidated_peaks = consolidated_peaks
@@ -210,3 +186,48 @@ class Project:
 		return ms_comparison_df
 
 		# chart_data = make_chart_data(self)
+
+
+def consolidate(
+		project: Project,
+		engine: pyms_nist_search.Engine,
+		) -> pandas.DataFrame:
+	"""
+	Consolidate the compound identification from the experiments into a single dataset.
+
+	:param project:
+	:param engine:
+
+	:returns: List of consolidated peaks and :class:`pandas.DataFrame`
+		giving the results of pairwise mass spectral comparisons between the repeats for each aligned peak.
+
+	.. versionadded:: 0.10.0
+	"""
+
+	ms_comparison_df = pairwise_ms_comparisons(project.alignment)
+
+	peak_numbers: List[int] = []
+	peak: Optional[QualifiedPeak]
+
+	qualified_peak_array = []
+
+	# for experiment in project.alignment.expr_code:
+	for experiment in project.datafile_data:
+		qualified_peaks = project.datafile_data[experiment].qualified_peaks
+		assert qualified_peaks is not None
+		for peak in qualified_peaks:
+			assert peak.peak_number is not None
+			peak_numbers.append(peak.peak_number)
+		qualified_peak_array.append(qualified_peaks)
+
+	# Convert peak_numbers to a set and sort smallest to largest
+	peak_numbers = sorted(set(peak_numbers))
+
+	consolidated_peaks = match_counter(
+			engine=engine,
+			peak_numbers=peak_numbers,
+			qualified_peaks=qualified_peak_array,
+			ms_comp_data=ms_comparison_df,
+			)
+
+	return consolidated_peaks, ms_comparison_df

libgunshotmatch/utils.py

@@ -32,6 +32,7 @@ from typing import TYPE_CHECKING, Any, Iterable, List, Optional, Sequence, Tuple
 
 # 3rd party
 import numpy
+from attr import AttrsInstance
 from chemistry_tools.spectrum_similarity import SpectrumSimilarity
 from mathematical.utils import rounders
 from pyms.DPA.Alignment import Alignment
@@ -114,10 +115,10 @@ def ms_comparison(top_ms: MassSpectrum, bottom_ms: MassSpectrum) -> Optional[flo
 	return match * 1000
 
 
-_O = TypeVar("_O", bound=object)
+_AI = TypeVar("_AI", bound=AttrsInstance)
 
 
-def _fix_init_annotations(method: Type[_O]) -> Type[_O]:
+def _fix_init_annotations(method: Type[_AI]) -> Type[_AI]:
 	init_annotations = method.__init__.__annotations__
 	cls_annotations = method.__annotations__
 

{libgunshotmatch-0.9.0.dist-info → libgunshotmatch-0.10.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: libgunshotmatch
-Version: 0.9.0
+Version: 0.10.0
 Summary: Base library for GunShotMatch.
 Author-email: Dominic Davis-Foster <dominic@davis-foster.co.uk>
 License: MIT
@@ -135,7 +135,7 @@ libgunshotmatch
 .. |language| image:: https://img.shields.io/github/languages/top/GunShotMatch/libgunshotmatch
 	:alt: GitHub top language
 
-.. |commits-since| image:: https://img.shields.io/github/commits-since/GunShotMatch/libgunshotmatch/v0.9.0
+.. |commits-since| image:: https://img.shields.io/github/commits-since/GunShotMatch/libgunshotmatch/v0.10.0
 	:target: https://github.com/GunShotMatch/libgunshotmatch/pulse
 	:alt: GitHub commits since tagged version
 

{libgunshotmatch-0.9.0.dist-info → libgunshotmatch-0.10.0.dist-info}/RECORD

@@ -2,21 +2,21 @@ libgunshotmatch/comparison/__init__.py,sha256=g-KDfzG-oyRf2aeXlMaJJ3aWfQtQkr5-bG
 libgunshotmatch/comparison/_utils.py,sha256=aZ6fnFM4M_vTsKq5t79V4tfven1eVdG4EUCuYFuXMVs,1895
 libgunshotmatch/comparison/projects.py,sha256=Dq294cDndmQI2FnIcVXwleCQF0mF2m8U5wWnO5_aD9A,3356
 libgunshotmatch/comparison/unknowns.py,sha256=kvJJDtfpMp-c3hjCOz7YtvuuDR0YksWmyUuaSBC_5Jg,2710
-libgunshotmatch/consolidate/__init__.py,sha256=11lblWDnQJAxd5Gz07KdhfGNYd5tYCL44PjCTAxowW4,22736
+libgunshotmatch/consolidate/__init__.py,sha256=zdnqx7YzV3MiW5IVFEiPSJZT43BdcOmlRqkIKOEBqPg,23816
 libgunshotmatch/consolidate/_fields.py,sha256=0kfPXJ0EG7GhdFiNzvcmd6W4i1x6Y0s2Y58z3RltPiA,2759
 libgunshotmatch/consolidate/_spectra.py,sha256=24aDoPwGWEyIFCH-fwRa4nifNMPqUyd-qTogk3BMeLY,2319
 libgunshotmatch/method/__init__.py,sha256=TP_3rvv3WEbV6Y5E_QWd2lcORPFnbXV4YRKGwXDa4ds,9423
 libgunshotmatch/method/_fields.py,sha256=HBFl0XmHBaAOHYdABs2NRbjtngGEw7kY0xP6D2Fl7h4,4941
-libgunshotmatch/__init__.py,sha256=aGZ1s8Ou1r0c75PysqNWM0LteNIEFNjLmhoAHHEvLZQ,1433
+libgunshotmatch/__init__.py,sha256=_8fk8Y_s16QckNbYhOXc8zxn6HIK0f_3TRYEIkYEYr4,1434
 libgunshotmatch/datafile.py,sha256=4C4BiR95PBfIUbiUKQzVfh57tgZu_l6rLS8bIgrgPUs,16246
-libgunshotmatch/gzip_util.py,sha256=pvD78_qnzRNCm4PK3KTb1cbOxzFfqTYcHB3VRi25oL0,2270
+libgunshotmatch/gzip_util.py,sha256=PcfT4QC4TM0KI3uCGetRpjChbTXTGmNEJBr7BghO3i8,2414
 libgunshotmatch/peak.py,sha256=huZ1hObIFcQpMPXyU6lYe7B-FFT-djSeG4jII7CgWAE,16923
-libgunshotmatch/project.py,sha256=rOPO1NAY5gxJrMSPhPXVMhBSvcsaplXxGnWRdo-fXDA,6360
+libgunshotmatch/project.py,sha256=ch8v8egvcULjYk7_40KywgM-_j3mf2tvRTzAsIC6Kp0,6904
 libgunshotmatch/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 libgunshotmatch/search.py,sha256=GMP7_DMnXFq3TG8AT1WARIn4ObCfOGfTDakpGeKzJYw,3123
-libgunshotmatch/utils.py,sha256=SpKO72YE4coY6OOqACpxwrYfH9xJzC2EhaKUfm4YT8k,5687
-libgunshotmatch-0.9.0.dist-info/LICENSE,sha256=bFtJt-lyVJHV-88FeFa_r3BEOsmpna5qG2KOl9JUNfU,1064
-libgunshotmatch-0.9.0.dist-info/METADATA,sha256=uYQCjLg9hFQe4Sc1PBF67Y0myIkT0_AI8rpF7ywT070,6392
-libgunshotmatch-0.9.0.dist-info/WHEEL,sha256=3vSnEhs48RUlSXcCXKCl4DOHs_qqaP2dU3IGkMqN2oI,84
-libgunshotmatch-0.9.0.dist-info/entry_points.txt,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-libgunshotmatch-0.9.0.dist-info/RECORD,,
+libgunshotmatch/utils.py,sha256=8wrdSk1ZWIgIu6eN9QEDF23goxN7z2VszxcDYZZ8ov0,5729
+libgunshotmatch-0.10.0.dist-info/LICENSE,sha256=bFtJt-lyVJHV-88FeFa_r3BEOsmpna5qG2KOl9JUNfU,1064
+libgunshotmatch-0.10.0.dist-info/METADATA,sha256=B26beZpWIdl-1YRu3Bl51SqsjWIOd0Y4vy2C1Osy7oA,6394
+libgunshotmatch-0.10.0.dist-info/WHEEL,sha256=3vSnEhs48RUlSXcCXKCl4DOHs_qqaP2dU3IGkMqN2oI,84
+libgunshotmatch-0.10.0.dist-info/entry_points.txt,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+libgunshotmatch-0.10.0.dist-info/RECORD,,