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_to_sirius__custom_candidate_db(self):
# Spectrum 1 -------------------
spec = MBSpectrum(os.path.join(os.path.dirname(__file__), "example_massbank_records", "FIO00665.txt"))
out = spec.to_sirius_format(
molecular_candidates=pd.read_csv(
os.path.join(os.path.dirname(__file__), "example_massbank_records", "FIO00665.tsv"), sep="\t"
)
)
self.assertIn("FIO00665.ms", out)
self.assertIn("FIO00665.tsv", out)
self.assertIsNotNone(out["FIO00665.tsv"])