def write_out_matrix_file(matrix_outfile, allruns, multipeptides, fraction_needed_selected,
style="none", write_requant=True, aligner_mscore_treshold=1.0):
matrix_writer = getwriter(matrix_outfile)
run_ids = [r.get_id() for r in allruns]
header = ["Peptide", "Protein"]
for r in allruns:
fname = "%s_%s" % (os.path.basename(r.orig_filename), r.get_id() )
header.extend(["Intensity_%s" % fname])
if style == "RT" or style == 'full':
header.extend(["RT_%s" % fname])
if style == "score" or style == 'full':
header.extend(["score_%s" % fname])
header.extend(["RT_mean", "RT_std", "pg_pvalue"])
for i in header:
matrix_writer.write(i)
matrix_writer.newline()
for multipep in multipeptides:
# Retrieve all transition group ids available for this precursor group
# Iterate through all ids and write one line per transition group id
trgr_ids = set([ trgr.get_id() for prgr in multipep.getPrecursorGroups() for trgr in prgr ])
for trgr_id in trgr_ids:
# Get all selected peakgroups that correspond to the current
# transition group id and ensure we do not have any twice.
allruns = [pg.peptide.run.get_id() for pg in multipep.get_selected_peakgroups()
if pg.peptide.get_id() == trgr_id]
if len(allruns) != len(set(allruns)):
# TODO test this as well ....
raise Exception("Error when writing out matrix, found more than one peakgroup for a run %s" % allruns)
# Get all selected peakgroups that correspond to the current transition group id
selected_peakgroups = dict([(pg.peptide.run.get_id(), pg)
for pg in multipep.get_selected_peakgroups() if pg.peptide.get_id() == trgr_id])
# Skip empty lines or lines that have too few entries
if len(selected_peakgroups) == 0:
continue
if (len(selected_peakgroups) * 1.0 / len(allruns) < fraction_needed_selected):
continue
# Write first two columns of the matrix
for i in [trgr_id, multipep.find_best_peptide_pg().peptide.protein_name]:
matrix_writer.write(i)
# Write other columns (one or two per run, depending on format)
rts = []
for rid in run_ids:
pg = selected_peakgroups.get(rid, None)
if not write_requant:
if not pg is None and pg.get_fdr_score() > 1.0:
pg = None
if pg is None:
matrix_writer.write("NA")
if style == "RT" or style == "full":
matrix_writer.write("NA")
if style == "score" or style == "full":
matrix_writer.write("NA")
else:
if pg.get_fdr_score() > 1.0:
color = 'r'
elif pg.get_fdr_score() > aligner_mscore_treshold:
color = 'b'
else:
color = 'd'
matrix_writer.write(pg.get_intensity(), color=color)
if style == "RT" or style == "full":
matrix_writer.write(pg.get_normalized_retentiontime(), color=color)
if style == "score" or style == "full":
matrix_writer.write(pg.get_fdr_score(), color=color)
if not pg is None:
rts.append(pg.get_normalized_retentiontime())
# The d_score is a z-score which computed on the null / decoy
# distribution which is (assumed) gaussian with u = 0, sigma = 1
# -> we thus compute a p-value from the z-score and assuming
# independent measurements, we multiply the p-values to compute a
# peakgroup p-value.
# We use norm.sf (1-cdf) on the vector of z-scores.
pvals = [float(pg.get_dscore()) for k,pg in selected_peakgroups.iteritems() if
not pg is None and not pg.get_dscore() is None]
pvalue = numpy.prod(scipy.stats.norm.sf(pvals))
for i in [numpy.mean(rts), numpy.std(rts), pvalue]:
matrix_writer.write(i)
matrix_writer.newline()
del matrix_writer
def write_out_matrix_file(matrix_outfile,
allruns,
multipeptides,
fraction_needed_selected,
style="none",
write_requant=True,
aligner_mscore_treshold=1.0):
matrix_writer = getwriter(matrix_outfile)
run_ids = [r.get_id() for r in allruns]
header = ["Peptide", "Protein"]
for r in allruns:
fname = "%s_%s" % (os.path.basename(r.orig_filename), r.get_id())
header.extend(["Intensity_%s" % fname])
if style == "RT" or style == 'full':
header.extend(["RT_%s" % fname])
if style == "score" or style == 'full':
header.extend(["score_%s" % fname])
header.extend(["RT_mean", "RT_std", "pg_pvalue"])
for i in header:
matrix_writer.write(i)
matrix_writer.newline()
for multipep in multipeptides:
# Retrieve all transition group ids available for this precursor group
# Iterate through all ids and write one line per transition group id
trgr_ids = set([
trgr.get_id() for prgr in multipep.getPrecursorGroups()
for trgr in prgr
])
for trgr_id in trgr_ids:
# Get all selected peakgroups that correspond to the current
# transition group id and ensure we do not have any twice.
allruns = [
pg.peptide.run.get_id()
for pg in multipep.get_selected_peakgroups()
if pg.peptide.get_id() == trgr_id
]
if len(allruns) != len(set(allruns)):
# TODO test this as well ....
raise Exception(
"Error when writing out matrix, found more than one peakgroup for a run %s"
% allruns)
# Get all selected peakgroups that correspond to the current transition group id
selected_peakgroups = dict([
(pg.peptide.run.get_id(), pg)
for pg in multipep.get_selected_peakgroups()
if pg.peptide.get_id() == trgr_id
])
# Skip empty lines or lines that have too few entries
if len(selected_peakgroups) == 0:
continue
if (len(selected_peakgroups) * 1.0 / len(allruns) <
fraction_needed_selected):
continue
# Write first two columns of the matrix
for i in [
trgr_id,
multipep.find_best_peptide_pg().peptide.protein_name
]:
matrix_writer.write(i)
# Write other columns (one or two per run, depending on format)
rts = []
for rid in run_ids:
pg = selected_peakgroups.get(rid, None)
if not write_requant:
if not pg is None and pg.get_fdr_score() > 1.0:
pg = None
if pg is None:
matrix_writer.write("NA")
if style == "RT" or style == "full":
matrix_writer.write("NA")
if style == "score" or style == "full":
matrix_writer.write("NA")
else:
if pg.get_fdr_score() > 1.0:
color = 'r'
elif pg.get_fdr_score() > aligner_mscore_treshold:
color = 'b'
else:
color = 'd'
matrix_writer.write(pg.get_intensity(), color=color)
if style == "RT" or style == "full":
matrix_writer.write(pg.get_normalized_retentiontime(),
color=color)
if style == "score" or style == "full":
matrix_writer.write(pg.get_fdr_score(), color=color)
if not pg is None:
rts.append(pg.get_normalized_retentiontime())
# The d_score is a z-score which computed on the null / decoy
# distribution which is (assumed) gaussian with u = 0, sigma = 1
# -> we thus compute a p-value from the z-score and assuming
# independent measurements, we multiply the p-values to compute a
# peakgroup p-value.
# We use norm.sf (1-cdf) on the vector of z-scores.
pvals = [
float(pg.get_dscore())
for k, pg in selected_peakgroups.iteritems()
if not pg is None and not pg.get_dscore() is None
]
pvalue = numpy.prod(scipy.stats.norm.sf(pvals))
for i in [numpy.mean(rts), numpy.std(rts), pvalue]:
matrix_writer.write(i)
matrix_writer.newline()
del matrix_writer