def diff(data1, data2):
"""
@summary: Compares two GCMS_data objects
@param data1: GCMS data set 1
@type data1: pyms.GCMS.Class.GCMS_data
@param data2: GCMS data set 2
@type data2: pyms.GCMS.Class.GCMS_data
@author: Qiao Wang
@author: Andrew Isaac
@author: Vladimir Likic
"""
# get time attributes
time_list1 = data1.get_time_list()
time_list2 = data2.get_time_list()
#
# First, check if two data sets have the same number of retention
# times.
#
if not len(time_list1) == len(time_list2):
print " -> The number of retention time points different."
print " First data set: %d time points" % (len(time_list1))
print " Second data set: %d time points" % (len(time_list2))
print " Data sets are different."
return
else:
time_rmsd = rmsd(time_list1, time_list2)
print " Data sets have the same number of time points."
print " Time RMSD: %.2e" % (time_rmsd)
#
# Second, check if each scan has the same number of m/z intensities
#
print " Checking for consistency in scan lengths ...",
sys.stdout.flush()
scan_list1 = data1.get_scan_list()
scan_list2 = data2.get_scan_list()
if not len(scan_list1) == len(scan_list2):
# since the number of rention times are the same, this indicated
# some unexpected problem with data
error("inconsistency in data detected")
N = len(scan_list1)
for ii in range(N):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
if len(mass_list1) != len(mass_list2):
print "\n Different number of points detected in scan no. %d" % (
ii)
print " Data sets are different."
return
print "OK"
#
# Third, if here, calculate the max RMSD for m/z and intensities
#
print " Calculating maximum RMSD for m/z values and intensities ...",
sys.stdout.flush()
max_mass_rmsd = 0.0
max_intensity_rmsd = 0.0
for ii in range(N):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
intensity_list1 = scan1.get_intensity_list()
intensity_list2 = scan2.get_intensity_list()
mass_rmsd = rmsd(mass_list1, mass_list2)
if mass_rmsd > max_mass_rmsd:
max_mass_rmsd = mass_rmsd
intensity_rmsd = rmsd(intensity_list1, intensity_list2)
if intensity_rmsd > max_intensity_rmsd:
max_intensity_rmsd = intensity_rmsd
print "\n Max m/z RMSD: %.2e" % (max_mass_rmsd)
print " Max intensity RMSD: %.2e" % (max_intensity_rmsd)
def diff(data1, data2):
"""
Compares two GCMS_data objects
:param data1: GCMS data set 1
:type data1: pyms.GCMS.Class.GCMS_data
:param data2: GCMS data set 2
:type data2: pyms.GCMS.Class.GCMS_data
:author: Qiao Wang
:author: Andrew Isaac
:author: Vladimir Likic
"""
# get time attributes
time_list1 = data1.get_time_list()
time_list2 = data2.get_time_list()
# First, check if two data sets have the same number of retention times.
if len(time_list1) != len(time_list2):
print(" The number of retention time points differ.")
print(f" First data set: {len(time_list1):d} time points")
print(f" Second data set: {len(time_list2):d} time points")
print(" Data sets are different.")
return
else:
time_rmsd = rmsd(time_list1, time_list2)
print(" Data sets have the same number of time points.")
print(f" Time RMSD: {time_rmsd:.2e}")
# Second, check if each scan has the same number of m/z intensities
print(" Checking for consistency in scan lengths ...", end='')
sys.stdout.flush()
scan_list1 = data1.get_scan_list()
scan_list2 = data2.get_scan_list()
if not len(scan_list1) == len(scan_list2):
# since the number of rention times are the same, this indicated
# some unexpected problem with data
raise ValueError("inconsistency in data detected")
for ii in range(len(scan_list1)):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
if len(mass_list1) != len(mass_list2):
print(f"\n Different number of points detected in scan no. {ii:d}")
print(" Data sets are different.")
return
print("OK")
# Third, if here, calculate the max RMSD for m/z and intensities
print(" Calculating maximum RMSD for m/z values and intensities ...",
end='')
sys.stdout.flush()
max_mass_rmsd = 0.0
max_intensity_rmsd = 0.0
for ii in range(len(scan_list1)):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
intensity_list1 = scan1.get_intensity_list()
intensity_list2 = scan2.get_intensity_list()
mass_rmsd = rmsd(mass_list1, mass_list2)
if mass_rmsd > max_mass_rmsd:
max_mass_rmsd = mass_rmsd
intensity_rmsd = rmsd(intensity_list1, intensity_list2)
if intensity_rmsd > max_intensity_rmsd:
max_intensity_rmsd = intensity_rmsd
print(f"\n Max m/z RMSD: {max_mass_rmsd:.2e}")
print(f" Max intensity RMSD: {max_intensity_rmsd:.2e}")
def diff(data1, data2):
"""
@summary: Compares two GCMS_data objects
@param data1: GCMS data set 1
@type data1: pyms.GCMS.Class.GCMS_data
@param data2: GCMS data set 2
@type data2: pyms.GCMS.Class.GCMS_data
@author: Qiao Wang
@author: Andrew Isaac
@author: Vladimir Likic
"""
# get time attributes
time_list1 = data1.get_time_list()
time_list2 = data2.get_time_list()
#
# First, check if two data sets have the same number of retention
# times.
#
if not len(time_list1) == len(time_list2):
print " -> The number of retention time points different."
print " First data set: %d time points" % (len(time_list1))
print " Second data set: %d time points" % (len(time_list2))
print " Data sets are different."
return
else:
time_rmsd = rmsd(time_list1, time_list2)
print " Data sets have the same number of time points."
print " Time RMSD: %.2e" % ( time_rmsd )
#
# Second, check if each scan has the same number of m/z intensities
#
print " Checking for consistency in scan lengths ...",
sys.stdout.flush()
scan_list1 = data1.get_scan_list()
scan_list2 = data2.get_scan_list()
if not len(scan_list1) == len(scan_list2):
# since the number of rention times are the same, this indicated
# some unexpected problem with data
error("inconsistency in data detected")
N = len(scan_list1)
for ii in range(N):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
if len(mass_list1) != len(mass_list2):
print "\n Different number of points detected in scan no. %d" % ( ii )
print " Data sets are different."
return
print "OK"
#
# Third, if here, calculate the max RMSD for m/z and intensities
#
print " Calculating maximum RMSD for m/z values and intensities ...",
sys.stdout.flush()
max_mass_rmsd = 0.0
max_intensity_rmsd = 0.0
for ii in range(N):
scan1 = scan_list1[ii]
scan2 = scan_list2[ii]
mass_list1 = scan1.get_mass_list()
mass_list2 = scan2.get_mass_list()
intensity_list1 = scan1.get_intensity_list()
intensity_list2 = scan2.get_intensity_list()
mass_rmsd = rmsd(mass_list1, mass_list2)
if mass_rmsd > max_mass_rmsd:
max_mass_rmsd = mass_rmsd
intensity_rmsd = rmsd(intensity_list1, intensity_list2)
if intensity_rmsd > max_intensity_rmsd:
max_intensity_rmsd = intensity_rmsd
print "\n Max m/z RMSD: %.2e" % ( max_mass_rmsd )
print " Max intensity RMSD: %.2e" % ( max_intensity_rmsd )