def test_to_sirius(self):
# Spectrum 1 -------------------
spec = MBSpectrum(os.path.join(os.path.dirname(__file__), "example_massbank_records", "FIO00665.txt"))
out = spec.to_sirius_format()
self.assertIn("FIO00665.ms", out)
self.assertIn("FIO00665.tsv", out)
self.assertIsNone(out["FIO00665.tsv"])
self.assertIn(">profile qtof", out["FIO00665.ms"])
# check fragmentation peaks
tmp = out["FIO00665.ms"].split("\n")
for idx, _peak in enumerate(spec.get_peaks(), start=tmp.index(">ms2merged") + 1):
_mz, _int = tmp[idx].split(" ")
self.assertEqual(_peak, (float(_mz), float(_int)))
# Spectrum 2 -------------------
spec = MBSpectrum(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EQ308406.txt"))
out = spec.to_sirius_format()
self.assertIn("EQ308406.ms", out)
self.assertIn("EQ308406.tsv", out)
self.assertIsNone(out["EQ308406.tsv"])
self.assertIn(">profile orbitrap", out["EQ308406.ms"])
# check fragmentation peaks
tmp = out["EQ308406.ms"].split("\n")
for idx, _peak in enumerate(spec.get_peaks(), start=tmp.index(">ms2merged") + 1):
_mz, _int = tmp[idx].split(" ")
self.assertEqual(_peak, (float(_mz), float(_int)))
# Spectrum 3 --------------------
spectra = []
acc = []
spec_cnt = 0
original_accessions = [
'EA000412', 'EA000414', 'EA000401', 'EA000413', 'EA000408', 'EA000409', 'EA000405', 'EA000402', 'EA000406',
'EA000404', 'EA000411', 'EA000403', 'EA000407', 'EA000410'
]
for oacc in original_accessions:
mb_fn = os.path.join(os.path.dirname(__file__), "example_massbank_records", "%s.txt" % oacc)
spectra.append(MBSpectrum(mb_fn))
acc.append(spectra[-1].get("accession"))
spec_cnt += 1
self.assertIn(
">ms2merged",
MBSpectrum.merge_spectra(spectra, merge_peak_lists=True).to_sirius_format()["EA33002987.ms"]
)
self.assertEqual(
spec_cnt,
MBSpectrum.merge_spectra(spectra, merge_peak_lists=False).to_sirius_format()["EA33002987.ms"].count(">ms2peaks")
)
def test_spectra_merging__EAX000402(self):
"""
We compare our spectra merging with the strategy applied in [1] and originally proposed in [2].
References:
[1] "MetFrag relaunched: incorporating strategies beyond in silico fragmentation" by Ruttkies et al. (2016)
[2] "Alignment of high resolution mass spectra: development of a heuristic approach for metabolomics" by
Kazmi et al. (2006)
"""
# self.skipTest("Intensities are wired in the original file: EAX000402")
# Load the list of spectra to merge: EA0004[56][0-9].txt --> EAX000402.txt
spectra = []
for mb_fn in glob.iglob(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EA0004[56][0-9].txt")):
spectra.append(MBSpectrum(mb_fn))
# Run the spectra merging using hierarchical clustering
merged_spectrum = MBSpectrum.merge_spectra(spectra) # type: MBSpectrum
# Merged spectrum as used by [1]
peaks_ref = [
(117.0347000, 999),
(186.0677308, 999)
]
mzs_ref = list(zip(*peaks_ref))[0]
ints_ref = np.array(list(zip(*peaks_ref))[1])
np.testing.assert_almost_equal(mzs_ref, merged_spectrum.get_mz())
np.testing.assert_almost_equal(ints_ref / 999, merged_spectrum.get_int())
def test_metainformation_merging__FIO00665(self):
# Apply merge function to a single spectrum
mb_fn = os.path.join(os.path.dirname(__file__), "example_massbank_records", "FIO00665.txt")
spec = MBSpectrum(mb_fn)
spectra = [spec]
acc_ref = [os.path.basename(mb_fn).split(".")[0]]
rt_ref = spec.get("retention_time")
precmz_ref = spec.get("precursor_mz")
recordtitle_ref = spec.get("record_title")
ce_ref = spec.get("collision_energy")
# -------------------
# WITH RT AGGREGATION
# -------------------
merged_spectrum = MBSpectrum.merge_spectra(spectra, rt_agg_fun=np.min) # type: MBSpectrum
self.assertEqual("FBZONXHGGPHHIY-UHFFFAOYSA-N", merged_spectrum.get("inchikey"))
self.assertEqual(acc_ref, merged_spectrum.get("original_accessions"))
self.assertEqual("FIO", merged_spectrum.get("accession")[:3])
self.assertEqual(MBSpectrum._get_new_accession_id(merged_spectrum.get("original_accessions")),
merged_spectrum.get("accession"))
self.assertEqual(rt_ref, merged_spectrum.get("retention_time"))
self.assertEqual(precmz_ref, merged_spectrum.get("precursor_mz"))
self.assertEqual(recordtitle_ref, merged_spectrum.get("record_title"))
self.assertEqual([ce_ref], merged_spectrum.get("collision_energy"))
def test_spectra_merging__EAX281502(self):
"""
We compare our spectra merging with the strategy applied in [1] and originally proposed in [2].
References:
[1] "MetFrag relaunched: incorporating strategies beyond in silico fragmentation" by Ruttkies et al. (2016)
[2] "Alignment of high resolution mass spectra: development of a heuristic approach for metabolomics" by
Kazmi et al. (2006)
"""
# Load the list of spectra to merge: EA2815[56][0-9].txt --> EAX281502.txt
spectra = []
for mb_fn in glob.iglob(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EA2815[56][0-9].txt")):
spectra.append(MBSpectrum(mb_fn))
# Run the spectra merging using hierarchical clustering
merged_spectrum = MBSpectrum.merge_spectra(spectra) # type: MBSpectrum
# Merged spectrum as used by [1]
peaks_ref = [
(81.0220667, 257.0000000),
(96.0095000, 226.0000000),
(109.0170500, 259.0000000),
(111.0198667, 290.0000000),
(116.0504625, 559.0000000),
(118.0663111, 999.0000000),
(126.1288286, 348.0000000),
(156.0821000, 84.0000000),
(170.1188000, 2.0000000),
(172.0770500, 54.0000000),
(174.0561667, 114.0000000),
(182.1189250, 95.0000000),
(197.1296571, 173.0000000),
(200.0717900, 999.0000000),
(227.1402000, 67.0000000),
(230.1551000, 290.0000000),
(244.0616500, 15.0000000),
(257.2023400, 218.0000000),
(273.1971500, 176.0000000),
(276.0872500, 3.0000000),
(283.1819500, 70.0000000),
(285.1611800, 381.0000000),
(301.1921875, 459.0000000),
(317.1871200, 142.0000000),
(327.1710000, 128.0000000),
(331.2034500, 3.0000000),
(345.1821167, 999.0000000),
(359.1974333, 177.0000000),
(377.2083750, 999.0000000)
]
mzs_ref = list(zip(*peaks_ref))[0]
ints_ref = np.array(list(zip(*peaks_ref))[1])
np.testing.assert_almost_equal(mzs_ref, merged_spectrum.get_mz())
np.testing.assert_almost_equal(ints_ref / 999, merged_spectrum.get_int(), decimal=3)
def test_metainformation_merging__EAX000401(self):
# Load the list of spectra to merge: EA0004[01][0-9].txt --> EAX000401.txt
spectra = []
acc_ref = []
rt_ref = []
precmz_ref = []
recordtitle_ref = []
ce_ref = []
for mb_fn in glob.iglob(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EA0004[01][0-9].txt")):
spectra.append(MBSpectrum(mb_fn))
# collect some reference meta-information
acc_ref.append(os.path.basename(mb_fn).split(".")[0])
rt_ref.append(spectra[-1].get("retention_time"))
precmz_ref.append(spectra[-1].get("precursor_mz"))
recordtitle_ref.append(spectra[-1].get("record_title"))
ce_ref.append(spectra[-1].get("collision_energy"))
# -------------------
# WITH RT AGGREGATION
# -------------------
merged_spectrum = MBSpectrum.merge_spectra(spectra, rt_agg_fun=np.mean) # type: MBSpectrum
self.assertEqual("OUSYWCQYMPDAEO-UHFFFAOYSA-N", merged_spectrum.get("inchikey"))
self.assertEqual(acc_ref, merged_spectrum.get("original_accessions"))
self.assertEqual("EA", merged_spectrum.get("accession")[:2])
self.assertEqual(MBSpectrum._get_new_accession_id(merged_spectrum.get("original_accessions")),
merged_spectrum.get("accession"))
self.assertEqual("min", merged_spectrum.get("retention_time_unit"))
self.assertEqual(np.mean(rt_ref), merged_spectrum.get("retention_time"))
self.assertEqual(precmz_ref[0], merged_spectrum.get("precursor_mz"))
self.assertEqual(recordtitle_ref, merged_spectrum.get("record_title"))
self.assertEqual(ce_ref, merged_spectrum.get("collision_energy"))
# ----------------------
# WITHOUT RT AGGREGATION
# ----------------------
merged_spectrum = MBSpectrum.merge_spectra(spectra, rt_agg_fun=None) # type: MBSpectrum
self.assertEqual("min", merged_spectrum.get("retention_time_unit"))
self.assertEqual(rt_ref, merged_spectrum.get("retention_time"))
def test_spectra_merging__EAX000401(self):
"""
We compare our spectra merging with the strategy applied in [1] and originally proposed in [2].
References:
[1] "MetFrag relaunched: incorporating strategies beyond in silico fragmentation" by Ruttkies et al. (2016)
[2] "Alignment of high resolution mass spectra: development of a heuristic approach for metabolomics" by
Kazmi et al. (2006)
"""
# Load the list of spectra to merge: EA0004[01][0-9].txt --> EAX000401.txt
spectra = []
for mb_fn in glob.iglob(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EA0004[01][0-9].txt")):
spectra.append(MBSpectrum(mb_fn))
# Run the spectra merging using hierarchical clustering
merged_spectrum = MBSpectrum.merge_spectra(spectra) # type: MBSpectrum
# Merged spectrum as used by [1]
peaks_ref = [
(53.03852, 55),
(57.0447285714286, 93),
(65.0386, 116),
(77.0386, 884),
(81.03345, 12),
(85.0396545454545, 17),
(91.0542714285714, 123),
(92.0494875, 377),
(95.04925, 112),
(103.041733333333, 15),
(104.049541666667, 999),
(105.044757142857, 271),
(105.069975, 12),
(110.060033333333, 7),
(119.060475, 631),
(130.04005, 11),
(130.0652, 49),
(131.07295, 24),
(142.0652, 9),
(147.0554, 3),
(160.087069230769, 999),
(188.082038461538, 999)
]
mzs_ref = list(zip(*peaks_ref))[0]
ints_ref = np.array(list(zip(*peaks_ref))[1])
np.testing.assert_almost_equal(mzs_ref, merged_spectrum.get_mz())
np.testing.assert_almost_equal(ints_ref / 999, merged_spectrum.get_int(), decimal=3)
def test_to_metfrag(self):
# Spectrum 1 --------------------
out = MBSpectrum(os.path.join(os.path.dirname(__file__), "example_massbank_records", "FIO00665.txt")).to_metfrag_format(
**{"MetFragScoreWeights": [0.8, 0.2],
"MetFragScoreTypes": ["FragmenterScore", "PubChemNumberPatents"],
"LocalDatabasePath": "/path/to/db",
"ResultsPath": "/path/to/results",
"NumberThreads": 4,
"PeakListPath": "/path/to/peaks"}
)
self.assertIn("FIO00665.peaks", out)
self.assertIn("FIO00665.conf", out)
self.assertIn("PeakListPath=/path/to/peaks/FIO00665.peaks", out["FIO00665.conf"])
self.assertIn("MetFragScoreWeights=0.8,0.2\n", out["FIO00665.conf"])
self.assertIn("MetFragScoreTypes=FragmenterScore,PubChemNumberPatents\n", out["FIO00665.conf"])
self.assertIn("PrecursorIonMode=-1\n", out["FIO00665.conf"])
self.assertIn("IsPositiveIonMode=False\n", out["FIO00665.conf"])
# Spectrum 2 --------------------
out = MBSpectrum(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EQ308406.txt")).to_metfrag_format(
**{"MetFragScoreWeights": [1.0],
"MetFragScoreTypes": ["FragmenterScore"],
"LocalDatabasePath": "/path/to/db",
"ResultsPath": "/path/to/results",
"NumberThreads": 4,
"PeakListPath": "/path/to/peaks"}
)
self.assertIn("EQ308406.peaks", out)
self.assertIn("EQ308406.conf", out)
self.assertIn("PeakListPath=/path/to/peaks/EQ308406.peaks", out["EQ308406.conf"])
self.assertIn("MetFragScoreWeights=1.0\n", out["EQ308406.conf"])
self.assertIn("MetFragScoreTypes=FragmenterScore\n", out["EQ308406.conf"])
self.assertIn("PrecursorIonMode=1\n", out["EQ308406.conf"])
self.assertIn("IsPositiveIonMode=True\n", out["EQ308406.conf"])
# Spectrum 3 --------------------
spectra = []
acc = []
for mb_fn in glob.iglob(os.path.join(os.path.dirname(__file__), "example_massbank_records", "EA0004[01][0-9].txt")):
spectra.append(MBSpectrum(mb_fn))
acc.append(spectra[-1].get("accession"))
merged_spectrum = MBSpectrum.merge_spectra(spectra)
out = merged_spectrum.to_metfrag_format(
**{"MetFragScoreWeights": [1.0],
"MetFragScoreTypes": ["FragmenterScore"],
"LocalDatabasePath": "/path/to/db",
"ResultsPath": "/path/to/results",
"NumberThreads": 4,
"PeakListPath": "/path/to/peaks"}
)
peaks_fn = merged_spectrum.get("accession") + ".peaks"
config_fn = merged_spectrum.get("accession") + ".conf"
self.assertIn("PeakListPath=/path/to/peaks/" + peaks_fn, out[config_fn])
self.assertIn("MetFragScoreWeights=1.0\n", out[config_fn])
self.assertIn("MetFragScoreTypes=FragmenterScore\n", out[config_fn])
self.assertIn("PrecursorIonMode=1\n", out[config_fn])
self.assertIn("IsPositiveIonMode=True\n", out[config_fn])