aparamon
commented
6 years ago Validator script:
import sys
from indigo import *
indigo = Indigo()
for mol in indigo.iterateSDFile(sys.argv[1]):
flags = 0
prevmz = None
for peak in mol.getProperty('MASS SPECTRAL PEAKS').split('\n'):
mz, intens = peak.split()
mz = float(mz)
#if prevmz and mz < prevmz:
# print('{}: M/z={} Da out of order.'.format(mol.getProperty('ID'), mz),
# file=sys.stderr)
# flags += 1
if prevmz and mz == prevmz:
print('{}: Duplicate m/z={} Da.'.format(mol.getProperty('ID'), mz),
file=sys.stderr)
flags += 1
if prevmz and mz > 10*prevmz:
print('{}: M/z jump at m/z={} Da.'.format(mol.getProperty('ID'), mz),
file=sys.stderr)
flags += 1
prevmz = mz
if flags:
print(mol.getProperty('ID'))
print(file=sys.stderr)
takaakin
sneumann
meier-rene
Reporting upstream as suggested in https://bitbucket.org/fiehnlab/mona/issues/209/incorrect-m-z-values
Apparently, multiple MoNA GC/MS spectra have incorrect entries in MASS SPECTRAL PEAKS.
To reproduce: http://mona.fiehnlab.ucdavis.edu/spectra/display/JP011674
Spectrum window displays multiple peaks with m/z > 1000 Da, which is definitely out-of-range (molecule mass is 436.2438).
Closer examination reveals that MASS SPECTRAL PEAKS contains multiple entries which are out of order, and exceed the previous entry by factor of >10:
Luckily, all these should be easy to fix, by removing the extra trailing zeroes. (Were these markers of some sort the operator failed to remove before submitting, or OCR bugs?)
In attached list.txt is a list of all automatically-flagged suspicions records; more detailed information is available in log.txt.
Please note that there are some false positives due to rare strange ordering of MASS SPECTRAL PEAKS items: http://mona.fiehnlab.ucdavis.edu/spectra/display/JP011672 (what natsort?!)
However, only <400 records have non-conventional ordering, so it might be feasible to review them all.