def testrun(scorers_list=[],
scorers_settings=[],
batchindex=-1,
header='',
use_metalikeness=False,
use_formula=False,
use_elements=False):
#for batchindex in range(5):
logit('Assessing Fragprint based search. Batch %s...' % batchindex)
total_peaks = 0
best_results_neg = [0] * max_results_per_query
worst_results_neg = [0] * max_results_per_query
return_number_neg = [0] * max_results_per_query
neg_mode_count = 0
best_results_pos = [0] * max_results_per_query
worst_results_pos = [0] * max_results_per_query
return_number_pos = [0] * max_results_per_query
pos_mode_count = 0
logtime()
logit('Assuming correct adduct info and 0 isotope.')
logit('No formula assumption. Assuming Mass precision: %s ppm' % ppm)
logit('Only considering [M+H]+ and [M-H]- adducts for now')
peaks = []
for spectrum_index in range(len(specmanager.ms_spectra)):
print('%s of %s spectra preprocessed' %
(spectrum_index, len(specmanager.ms_spectra)))
spectrum = specmanager.ms_spectra[spectrum_index]
if batchindex == -1 or int(
spectrum.parameters['crossvalidation_batch_index']
) == batchindex:
for peak in spectrum.peaks:
if 'ion_type' in peak.parameters:
if peak.parameters[
'ion_type'] in FragPrintScorer.supported_adducts:
peak.ppm = ppm
#Assuming 0 isotope states only.
shortinchi = get_short_inchi_from_full_inchi(
peak.parent_spectrum.parameters['inchi'])
#
filters = []
if use_metalikeness:
metabolite_likeness = LikenessFilter(
likeness=['MetaLike'])
filters.append(metabolite_likeness)
if use_formula:
formula = peak.parent_spectrum.parameters[
'formula']
correct_formula = FormulasFilter(formula)
filters.append(correct_formula)
if use_elements:
elements = peak.parent_spectrum.parameters[
'formula']
correct_elements = ElementCompositionFilter(
elements, elements)
filters.append(correct_elements)
annotation=MSPeakAnnotation(peak, adduct=get_adduct_by_name(peak.parameters['ion_type'],\
spectrum.parameters['mode']), isotope=0, formula_scorer=None,\
filters=filters, scores={'AdductIsotopeScore':1.0})
if not (annotation.adduct is None):
peak.annotations = [annotation]
peaks.append(peak)
else:
logit('Unsupported Adduct: %s' %
peak.parameters['ion_type'])
logit('Peaks to annotate: %s' % len(peaks))
#
peak_annotator.annotate_peaks(peaks, test_chemical_databases, \
scorers_list=scorers_list, scorers_settings=scorers_settings, total_score=total_multiplicative_score, required_fields=set(['ShortInChI']),\
results_limit=max_results_per_query, save_memory=False, ppm=ppm, overwrite=True)
for i in reversed(range(len(peaks))):
if peaks[i].annotations[0].mol_candidates is None:
del peaks[i]
logit('Finished annotating... Annotated peaks :%s' % len(peaks))
logtime()
total_peaks = len(peaks)
total_candidates = 0
for peakindex in range(total_peaks):
peak = peaks[peakindex]
peakmode = 0
if peak.parent_spectrum.parameters['mode'] == 1:
pos_mode_count += 1
peakmode = 1
elif peak.parent_spectrum.parameters['mode'] == -1:
neg_mode_count += 1
peakmode = -1
#print('%s of %s '%(peakindex,total_peaks));
shortinchi = get_short_inchi_from_full_inchi(
peak.parent_spectrum.parameters['inchi'])
for annotation in peak.annotations:
annotation.min_correct = -1
annotation.max_correct = -1
total_candidates += annotation.mol_candidates.total_candidate_count
#Get annotation score
annotation.mean_score = 0.0
annotation.max_score = 0.0
for index in range(len(annotation.mol_candidates.mol_list)):
total_score = annotation.mol_candidates.mol_list[index][
'TotalScore']
#print(annotation.mol_candidates.mol_list[index]);
annotation.mean_score += total_score
if total_score > annotation.max_score:
annotation.max_score = total_score
if len(annotation.mol_candidates.mol_list) > 0:
annotation.mean_score = annotation.mean_score / len(
annotation.mol_candidates.mol_list)
if len(annotation.mol_candidates.mol_list
) <= max_results_per_query:
if peakmode == 1:
return_number_pos[
len(annotation.mol_candidates.mol_list) -
1] += 1
elif peakmode == -1:
return_number_neg[
len(annotation.mol_candidates.mol_list) -
1] += 1
for index in reversed(
range(1, len(annotation.mol_candidates.mol_list))):
if annotation.mol_candidates.mol_list[index - 1][
'ShortInChI'] == annotation.mol_candidates.mol_list[
index]['ShortInChI']:
del annotation.mol_candidates.mol_list[index]
# For the purpose of statistics condensing sequential identical Short InChI-s
#print(len(annotation.mol_candidates.mol_list));
#print('Correct: %s'%shortinchi);
for index in range(len(annotation.mol_candidates.mol_list)):
#print(annotation.mol_candidates.mol_list[index]['ShortInChI']);
if annotation.mol_candidates.mol_list[index][
'ShortInChI'] == shortinchi:
#print('Got it!');
if annotation.min_correct == -1:
annotation.min_correct = index
annotation.max_correct = index
if annotation.min_correct > -1:
if peakmode == 1:
for i in range(len(best_results_pos)):
if annotation.min_correct <= i:
best_results_pos[i] += 1
if annotation.max_correct <= i:
worst_results_pos[i] += 1
elif peakmode == -1:
for i in range(len(best_results_neg)):
if annotation.min_correct <= i:
best_results_neg[i] += 1
if annotation.max_correct <= i:
worst_results_neg[i] += 1
#else:
# logit('No candidate for %s (Mass: %s, peak mode: %s, peak_mz %s, deltaM %s)'%(shortinchi,peak.parent_spectrum.parameters['exactmass'],peakmode, peak.mz,abs(peak.mz-float(peak.parent_spectrum.parameters['exactmass']))));
logit('Finished Calculating Retrieval Stats.')
logit('Positive Mode Count: %s' % pos_mode_count)
logit('Negative Mode Count: %s' % neg_mode_count)
logit('Total Peaks: %s' % total_peaks)
logit('Total Candidate Count: %s' % total_candidates)
logtime()
logit('Positive Mode (Total: %s ):' % pos_mode_count)
if pos_mode_count > 0:
for i in range(max_results_per_query):
logit(
'Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'
% (i + 1, best_results_pos[i] * 100 / pos_mode_count,
worst_results_pos[i] * 100 / pos_mode_count,
return_number_pos[i]))
logit('Negative Mode (Total: %s ):' % neg_mode_count)
if neg_mode_count > 0:
for i in range(max_results_per_query):
logit(
'Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'
% (i + 1, best_results_neg[i] * 100 / neg_mode_count,
worst_results_neg[i] * 100 / neg_mode_count,
return_number_neg[i]))
logit('Both Modes (Total: %s ):' % (pos_mode_count + neg_mode_count))
if neg_mode_count > 0 or pos_mode_count > 0:
for i in range(max_results_per_query):
logit(
'Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'
% (i + 1,
(best_results_neg[i] + best_results_pos[i]) * 100 /
(neg_mode_count + pos_mode_count),
(worst_results_neg[i] + worst_results_pos[i]) * 100 /
(neg_mode_count + pos_mode_count),
return_number_pos[i] + return_number_neg[i]))
for i in range(len(best_results_neg)):
best_results_neg[i] = best_results_neg[i] * 100 / neg_mode_count
worst_results_neg[i] = worst_results_neg[i] * 100 / neg_mode_count
best_results_pos[i] = best_results_pos[i] * 100 / pos_mode_count
worst_results_pos[i] = worst_results_pos[i] * 100 / pos_mode_count
store_stats(header, best_results_neg, worst_results_neg,
best_results_pos, worst_results_pos)
#logit('Missed Peaks due to unknown mode: %s'%(total_peaks-neg_mode_count-pos_mode_count));
logtime()
'''
def run_annotation(spectral_input_path, db_file, chemical_databases, output_folder, SVMs_path, ncpu, ppm, max_results_per_query, test_mode=False, ignore_peaks_without_spectra=True):
global starttime;
global logfile;
if not os.path.isfile(db_file):
print('No database list file found!');
quit();
db_files=[db_file];
if not os.path.exists(output_folder):
try:
os.makedirs(output_folder);
except:
print('Cannot create output path %s !'%output_folder);
print('Run: python annotate.py -h for help');
sys.exit(2);
logfile=open(os.path.join(output_folder,'log.txt'),'w');
starttime=time.time();
logit('Starting. %s'%time.strftime('%d/%m/%y %H:%M:%S'));
logit('Input Spectral path: %s'%spectral_input_path);
logit('Output path: %s'%output_folder)
if chemical_databases:
logit('Chemical Databases Included: %s'%chemical_databases);
else:
logit('Chemical Databases Included: All');
logit('from chemical database paths file %s'%db_file);
logit('SVMs from %s'%SVMs_path);
logit('nCPUs to use: %s'%ncpu);
logit('Default ppm tolerance for MS1: %s'%ppm);
specmanager=SpectralManager();
logit('Reading test spectra...');
specmanager.import_textfile_spectra_from_folder(spectral_input_path);
logit('Finshed reading spectra. Total number: %s'%len(specmanager.ms_spectra));
logtime();
multiprocessing.freeze_support();
scorers_list=['FingerPrintScorer','FragPrintScorer'];
scorers_settings=[[],[]];
total_score=total_multiplicative_score;
with PeakAnnotator(db_files, SVMs_path, ncpu) as peak_annotator:
logit('Beginning annotation...');
logtime();
total_peaks=0;
best_results_neg=[0]*max_results_per_query;
worst_results_neg=[0]*max_results_per_query;
return_number_neg=[0]*max_results_per_query;
neg_mode_count=0;
best_results_pos=[0]*max_results_per_query;
worst_results_pos=[0]*max_results_per_query;
return_number_pos=[0]*max_results_per_query;
pos_mode_count=0;
peaks=[];
supported_adducts=set(global_supported_adducts);
#Collect all supported adducts
#if 'FingerPrintScorer' in scorers_list:
# supported_adducts=supported_adducts|FingerPrintScorer.supported_adducts;
#if 'FragPrintScorer' in scorers_list:
# supported_adducts=supported_adducts|FragPrintScorer.supported_adducts;
#Leave only adducts supported by all selected filters
if 'FingerPrintScorer' in scorers_list:
supported_adducts=supported_adducts&FingerPrintScorer.supported_adducts;
if 'FragPrintScorer' in scorers_list:
supported_adducts=supported_adducts&FragPrintScorer.supported_adducts;
logit('Preparing annotations');
if test_mode:
logit('Running in test mode.....');
logit('Assuming correct adduct info and 0 isotope.');
logit('No formula assumption. Mass precision: %s ppm'%ppm);
logit('Only considering [M+H]+ and [M-H]- adducts for now');
print('\n');
for spectrum_index in range(len(specmanager.ms_spectra)):
#print('\r%s of %s spectra preprocessed'%(spectrum_index,len(specmanager.ms_spectra)));
spectrum=specmanager.ms_spectra[spectrum_index];
for peak in spectrum.peaks:
if (not ignore_peaks_without_spectra) or (hasattr(peak,'ms_spectra') and peak.ms_spectra):
if hasattr(peak,'parameters') and ('ion_type' in peak.parameters):
if (peak.parameters['ion_type'] in FragPrintScorer.supported_adducts):
peak.ppm=ppm;
shortinchi=get_short_inchi_from_full_inchi(peak.parent_spectrum.parameters['inchi']);
annotation=MSPeakAnnotation(peak, adduct=\
get_adduct_by_name(peak.parameters['ion_type'], \
spectrum.parameters['mode']), isotope=0, \
formula_scorer=None, element_scorer=None, filters=[], scores={});
#testformulascorer=FormulaScorer();
#testformulascorer.setup_scorer(peak.parent_spectrum.parameters['formula'],1.0);
#testelementscorer=ElementScorer();
#testelementscorer.setup_scorer({'C':0.8,'O':0.9,'Si':0.1});
#formulafilter=FormulasFilter(formulas=[peak.parent_spectrum.parameters['formula']]);
#elementfilter=ElementCompositionFilter(peak.parent_spectrum.parameters['formula'], peak.parent_spectrum.parameters['formula']);
#inchifilter=InChIFilter(ref_inchi=peak.parent_spectrum.parameters['inchi'],use_short_inchi=False, match_type=0)
#annotation=MSPeakAnnotation(peak, adduct=\
# get_adduct_by_name(peak.parameters['ion_type'], \
# spectrum.parameters['mode']), isotope=0, \
# formula_scorer=testformulascorer, element_scorer=testelementscorer, \
# filters=[formulafilter, inchifilter, elementfilter], scores={'ExtraScore':25.0, 'AllExtra':1.9});
if not (annotation.adduct is None):
peak.annotations=[annotation];
peaks.append(peak);
else:
logit('Unsupported Adduct: %s'%peak.parameters['ion_type']);
else:
logit('Running in normal mode.....');
logit('Mass precision: %s ppm'%ppm);
print('\n');
for spectrum_index in range(len(specmanager.ms_spectra)):
#print('\r%s of %s spectra preprocessed'%(spectrum_index,len(specmanager.ms_spectra)));
spectrum=specmanager.ms_spectra[spectrum_index];
for peak in spectrum.peaks:
if (not ignore_peaks_without_spectra) or (hasattr(peak,'ms_spectra') and peak.ms_spectra):
peak.ppm=ppm;
generate_annotation=True;
if hasattr(peak, 'annotations'):
if peak.annotations:
generate_annotation=False;
peaks.append(peak);
if generate_annotation:
#Assuming isotope 0 by default
if spectrum.parameters['mode']==1:
selected_adducts=get_positive_mode_adducts(supported_adducts);
else:
selected_adducts=get_negative_mode_adducts(supported_adducts);
peak.annotations=[];
for adduct in selected_adducts:
annotation=MSPeakAnnotation(peak, adduct=adduct, isotope=0, formula_scorer=None, element_scorer=None, filters=[], scores={});
peak.annotations.append(annotation);
peaks.append(peak);
logit('Peaks to annotate: %s'%len(peaks));
if test_mode:
required_fields=set(['ShortInChI','InChI','SMILES','Formula','IDs']);
#required_fields=set(['ShortInChI','InChI','SMILES','Formula','IDs', 'InChIKeyValues', 'InChiKey', 'FormulaVector', 'ElementVector', 'Frag', 'FPT']);
else:
required_fields=set(['InChI','SMILES','Formula','IDs']);
peak_annotator.annotate_peaks(peaks, chemical_databases, \
scorers_list=scorers_list, scorers_settings=scorers_settings, total_score=total_score, required_fields=required_fields,\
results_limit=max_results_per_query, save_memory=False, ppm=ppm, overwrite=True);
annotation_count=0;
total_candidates=0;
total_peaks=len(peaks);
for peakindex in range(total_peaks):
peak=peaks[peakindex];
if peak.annotations:
for annotation in peak.annotations:
if not (annotation.mol_candidates is None):
annotation_count+=1;
total_candidates+=annotation.mol_candidates.total_candidate_count;
logit('Finished annotating... ');
logit('Annotated peaks :%s'%total_peaks);
if total_peaks>0:
logit('Averaged annotations per peak:%s'%(float(annotation_count)/total_peaks));
logit('Averaged candidate molecules per peak:%s'%(float(total_candidates)/total_peaks));
#Calculating retrieval statistics using test data
#=====================================================================
if test_mode:
#logit('Removing peaks with no annotations found from consideration...');
#for i in reversed(range(len(peaks))):
# if not peaks[i].annotations:
# del peaks[i];
#logit('Finished removing... Annotated peaks :%s'%len(peaks));
total_peaks=len(peaks);
print('Calculating retrieval stats...\n');
for peakindex in range(total_peaks):
peak=peaks[peakindex];
peakmode=0;
if peak.parent_spectrum.parameters['mode']==1:
pos_mode_count+=1;
peakmode=1;
elif peak.parent_spectrum.parameters['mode']==-1:
neg_mode_count+=1;
peakmode=-1;
#print('\r%s of %s '%(peakindex,total_peaks));
shortinchi=get_short_inchi_from_full_inchi(peak.parent_spectrum.parameters['inchi']);
for annotation in peak.annotations:
annotation.min_correct=-1;
annotation.max_correct=-1;
annotation.mean_score=0.0;
annotation.max_score=0.0;
for index in range(len(annotation.mol_candidates.mol_list)):
total_score=annotation.mol_candidates.mol_list[index]['TotalScore'];
annotation.mean_score+=total_score;
if total_score>annotation.max_score:
annotation.max_score=total_score;
if len(annotation.mol_candidates.mol_list)>0:
annotation.mean_score=annotation.mean_score/len(annotation.mol_candidates.mol_list);
if len(annotation.mol_candidates.mol_list)<=max_results_per_query:
if peakmode==1:
return_number_pos[len(annotation.mol_candidates.mol_list)-1]+=1;
elif peakmode==-1:
return_number_neg[len(annotation.mol_candidates.mol_list)-1]+=1;
for index in reversed(range(1,len(annotation.mol_candidates.mol_list))):
if annotation.mol_candidates.mol_list[index-1]['ShortInChI']==annotation.mol_candidates.mol_list[index]['ShortInChI']:
del annotation.mol_candidates.mol_list[index]; # For the purpose of statistics condensing sequential identical Short InChI-s
for index in range(len(annotation.mol_candidates.mol_list)):
if annotation.mol_candidates.mol_list[index]['ShortInChI']==shortinchi:
annotation.mol_candidates.mol_list[index]['Annotation']='Correct';
if annotation.min_correct==-1:
annotation.min_correct=index;
annotation.max_correct=index;
else:
annotation.mol_candidates.mol_list[index]['Annotation']='Wrong';
if annotation.min_correct>-1:
if peakmode==1:
for i in range(len(best_results_pos)):
if annotation.min_correct<=i:
best_results_pos[i]+=1;
if annotation.max_correct<=i:
worst_results_pos[i]+=1;
elif peakmode==-1:
for i in range(len(best_results_neg)):
if annotation.min_correct<=i:
best_results_neg[i]+=1;
if annotation.max_correct<=i:
worst_results_neg[i]+=1;
logit('Finished Calculating Retrieval Stats.');
logit('Positive Mode Count: %s'%pos_mode_count);
logit('Negative Mode Count: %s'%neg_mode_count);
logit('Total Peaks: %s'%total_peaks);
logit('Positive Mode (Total: %s ):'%pos_mode_count);
if pos_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%.2f%%\tWorst:\t%.2f%%\tCCount:\t%s'%(i+1,float(best_results_pos[i])*100/pos_mode_count,worst_results_pos[i]*100/pos_mode_count,return_number_pos[i]));
logit('Negative Mode (Total: %s ):'%neg_mode_count);
if neg_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%.2f%%\tWorst:\t%.2f%%\tCCount:\t%s'%(i+1,float(best_results_neg[i])*100/neg_mode_count,worst_results_neg[i]*100/neg_mode_count,return_number_neg[i]));
logit('Both Modes (Total: %s ):'%(pos_mode_count+neg_mode_count));
if neg_mode_count>0 or pos_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%.2f%%\tWorst:\t%.2f%%\tCCount:\t%s'%(i+1,float(best_results_neg[i]+best_results_pos[i])*100/(neg_mode_count+pos_mode_count),float(worst_results_neg[i]+worst_results_pos[i])*100/(neg_mode_count+pos_mode_count),return_number_pos[i]+return_number_neg[i]));
for peak in peaks:
merge_annotations(peak, remove_old_annotations=False);
#Finished calculating retrieval statistics using test data
#=====================================================================
logtime();
logit('Finished Annotation. Exporting results...');
logtime();
logit('Exporting to JSON...');
spectra_to_json(os.path.join(output_folder,'annotated_spectra.json'), specmanager.ms_spectra);
logtime();
logit('Exporting to internal text format...');
specmanager.export_textfile_spectra_to_folder(os.path.join(output_folder,'annotated_spectra'));
logit('Preparing HTML report...');
generate_HTML_report(os.path.join(output_folder,'Report'), specmanager);
specmanager.close();
logtime();
logit('Finished');
logfile.close();
def _pipe_from_textfile(self, finp):
current_peak = None
current_peak_number = 0
while True:
s = finp.readline()
if s == '':
return
s = s.rstrip('\n').lstrip()
if '##' in s:
s = s[:s.index('##')]
if '=' in s:
s = s.split('=', 1)
if s[0].lower().startswith('mode'):
self.parameters['mode'] = int(s[1])
elif s[0].lower().startswith('collision_energy'):
self.parameters['collision_energy'] = float(s[1])
elif s[0].lower().startswith('level'):
self.parameters['level'] = int(s[1])
else:
self.parameters[s[0].lower()] = s[1]
elif s.lower().startswith('peaks'):
finished = False
while not finished:
s = finp.readline()
if s == '':
return
s = s.rstrip('\n').lstrip()
if '##' in s:
s = s[:s.index('##')]
if '=' in s:
s = s.split('=', 1)
if not hasattr(current_peak, 'parameters'):
if isinstance(current_peak, MSBasicPeak):
current_peak = MSPeak(current_peak)
current_peak.number = current_peak_number
current_peak.parent_spectrum = self
self.peaks[-1] = current_peak
current_peak.parameters = {}
current_peak.parameters[s[0].lower()] = s[1]
elif s.lower().startswith('spectrum'):
sub_spectrum = MSSpectrum()
sub_spectrum._pipe_from_textfile(finp)
if hasattr(current_peak, 'ms_spectra'):
current_peak.ms_spectra.append(sub_spectrum)
else:
if isinstance(current_peak, MSBasicPeak):
current_peak = MSPeak(current_peak)
current_peak.number = current_peak_number
current_peak.parent_spectrum = self
self.peaks[-1] = current_peak
current_peak.ms_spectra = [sub_spectrum]
sub_spectrum.parent_peak = current_peak
elif s.lower().startswith('annotation'):
annotation = MSPeakAnnotation(current_peak)
annotation._pipe_from_textfile(finp)
if hasattr(current_peak, 'annotations'):
current_peak.annotations.append(annotation)
else:
if isinstance(current_peak, MSBasicPeak):
current_peak = MSPeak(current_peak)
current_peak.number = current_peak_number
current_peak.parent_spectrum = self
self.peaks[-1] = current_peak
current_peak.annotations = [annotation]
elif s.lower().startswith('merged_annotation'):
merged_annotation = MolecularRecord()
merged_annotation._pipe_from_textfile(finp)
if hasattr(current_peak, 'merged_annotations'):
current_peak.merged_annotations.append(
merged_annotation)
else:
if isinstance(current_peak, MSBasicPeak):
current_peak = MSPeak(current_peak)
current_peak.number = current_peak_number
current_peak.parent_spectrum = self
self.peaks[-1] = current_peak
current_peak.merged_annotations = [
merged_annotation
]
elif (s.lower().startswith('end')):
finished = True
else:
s = s.split(',')
if self.parameters['level'] > 1:
current_peak = MSBasicPeak()
else:
current_peak = MSPeak()
current_peak.number = int(s[0])
current_peak.parent_spectrum = self
current_peak.mz = float(s[1])
current_peak.intensity = float(s[2])
current_peak_number = int(s[0])
if len(s) > 3:
if isinstance(current_peak, MSBasicPeak):
current_peak = MSPeak(current_peak)
current_peak.number = current_peak_number
current_peak.parent_spectrum = self
current_peak.ppm = float(s[3])
self.peaks.append(current_peak)
elif s.lower().startswith('end'):
return
def testrun(scorers_list=[], scorers_settings=[], batchindex=-1):
#for batchindex in range(5):
logit('Assessing Fragprint based search. Batch %s...'%batchindex);
total_peaks=0;
best_results_neg=[0]*max_results_per_query;
worst_results_neg=[0]*max_results_per_query;
return_number_neg=[0]*max_results_per_query;
neg_mode_count=0;
best_results_pos=[0]*max_results_per_query;
worst_results_pos=[0]*max_results_per_query;
return_number_pos=[0]*max_results_per_query;
pos_mode_count=0;
logtime();
logit('Assuming correct adduct info and 0 isotope.');
logit('No formula assumption. Assuming Mass precision: %s ppm'%ppm);
logit('Only considering [M+H]+ and [M-H]- adducts for now');
peaks=[];
for spectrum_index in range(len(specmanager.ms_spectra)):
print('%s of %s spectra preprocessed'%(spectrum_index,len(specmanager.ms_spectra)));
spectrum=specmanager.ms_spectra[spectrum_index];
if batchindex==-1 or int(spectrum.parameters['crossvalidation_batch_index'])==batchindex:
for peak in spectrum.peaks:
if 'ion_type' in peak.parameters:
if peak.parameters['ion_type'] in FragPrintScorer.supported_adducts:
peak.ppm=ppm; #Assuming 0 isotope states only.
shortinchi=peak.parent_spectrum.parameters['shortinchi'];
#formula=peak.parent_spectrum.parameters['formula'];
#correct_inchi=InChiFilter(shortinchi,True,4);
#correct_formula=FormulasFilter(formula);
annotation=MSPeakAnnotation(adduct=get_adduct_by_name(peak.parameters['ion_type'],\
spectrum.parameters['mode']), isotope=0, formula_scorer=None,\
filters=[], scores={'AdductIsotopeScore':1.0});
if not (annotation.adduct is None):
peak.annotations=[annotation];
annotation.parent_peak=peak;
peaks.append(peak);
else:
logit('Unsupported Adduct: %s'%peak.parameters['ion_type']);
logit('Peaks to annotate: %s'%len(peaks));
#
peak_annotator.annotate_peaks(peaks, test_chemical_databases, \
scorers_list=scorers_list, scorers_settings=scorers_settings, total_score=total_multiplicative_score,\
results_limit=max_results_per_query, save_memory=False, batch_index=batchindex, ppm=ppm, overwrite=True);
for i in reversed(range(len(peaks))):
if not peaks[i].annotations[0].mol_candidates:
del peaks[i];
logit('Finished annotating... Annotated peaks :%s'%len(peaks));
logtime();
total_peaks=len(peaks);
for peakindex in range(total_peaks):
peak=peaks[peakindex];
peakmode=0;
if peak.parent_spectrum.parameters['mode']==1:
pos_mode_count+=1;
peakmode=1;
elif peak.parent_spectrum.parameters['mode']==-1:
neg_mode_count+=1;
peakmode=-1;
print('%s of %s '%(peakindex,total_peaks));
shortinchi=peak.parent_spectrum.parameters['shortinchi'];
for annotation in peak.annotations:
annotation.min_correct=-1;
annotation.max_correct=-1;
#Get annotation score
annotation.mean_score=0.0;
annotation.max_score=0.0;
for index in range(len(annotation.mol_candidates.mol_list)):
total_score=annotation.mol_candidates.mol_list[index]['TotalScore'];
#print(annotation.mol_candidates.mol_list[index]);
annotation.mean_score+=total_score;
if total_score>annotation.max_score:
annotation.max_score=total_score;
if len(annotation.mol_candidates.mol_list)>0:
annotation.mean_score=annotation.mean_score/len(annotation.mol_candidates.mol_list);
if len(annotation.mol_candidates.mol_list)<=max_results_per_query:
if peakmode==1:
return_number_pos[len(annotation.mol_candidates.mol_list)-1]+=1;
elif peakmode==-1:
return_number_neg[len(annotation.mol_candidates.mol_list)-1]+=1;
for index in reversed(range(1,len(annotation.mol_candidates.mol_list))):
if annotation.mol_candidates.mol_list[index-1]['ShortInChi']==annotation.mol_candidates.mol_list[index]['ShortInChi']:
del annotation.mol_candidates.mol_list[index]; # For the purpose of statistics condensing sequential identical Short InChi-s
#print(len(annotation.mol_candidates.mol_list));
#print('Correct: %s'%shortinchi);
for index in range(len(annotation.mol_candidates.mol_list)):
#print(annotation.mol_candidates.mol_list[index]['ShortInChi']);
if annotation.mol_candidates.mol_list[index]['ShortInChi']==shortinchi:
#print('Got it!');
if annotation.min_correct==-1:
annotation.min_correct=index;
annotation.max_correct=index;
if annotation.min_correct>-1:
if peakmode==1:
for i in range(len(best_results_pos)):
if annotation.min_correct<=i:
best_results_pos[i]+=1;
if annotation.max_correct<=i:
worst_results_pos[i]+=1;
elif peakmode==-1:
for i in range(len(best_results_neg)):
if annotation.min_correct<=i:
best_results_neg[i]+=1;
if annotation.max_correct<=i:
worst_results_neg[i]+=1;
#else:
# logit('No candidate for %s (Mass: %s, peak mode: %s, peak_mz %s, deltaM %s)'%(shortinchi,peak.parent_spectrum.parameters['exactmass'],peakmode, peak.mz,abs(peak.mz-float(peak.parent_spectrum.parameters['exactmass']))));
logit('Finished Calculating Retrieval Stats.');
logit('Positive Mode Count: %s'%pos_mode_count);
logit('Negative Mode Count: %s'%neg_mode_count);
logit('Total Peaks: %s'%total_peaks);
logtime();
logit('Positive Mode (Total: %s ):'%pos_mode_count);
if pos_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'%(i+1,best_results_pos[i]*100/pos_mode_count,worst_results_pos[i]*100/pos_mode_count,return_number_pos[i]));
logit('Negative Mode (Total: %s ):'%neg_mode_count);
if neg_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'%(i+1,best_results_neg[i]*100/neg_mode_count,worst_results_neg[i]*100/neg_mode_count,return_number_neg[i]));
logit('Both Modes (Total: %s ):'%(pos_mode_count+neg_mode_count));
if neg_mode_count>0 or pos_mode_count>0:
for i in range(max_results_per_query):
logit('Correct within first\t%s:\tBest:\t%s%%\tWorst:\t%s%%\tCandidateCount:\t%s'%(i+1,(best_results_neg[i]+best_results_pos[i])*100/(neg_mode_count+pos_mode_count),(worst_results_neg[i]+worst_results_pos[i])*100/(neg_mode_count+pos_mode_count),return_number_pos[i]+return_number_neg[i]));
#logit('Missed Peaks due to unknown mode: %s'%(total_peaks-neg_mode_count-pos_mode_count));
logtime();
'''