def write_fdr_outputs(inFile, specFile, pepFile, protFile):
smf.print_milestone('Generating FDR Analysis Files:')
overallDf = pd.read_csv(inFile).sort_values(
'MaCC_Score', ascending=False).reset_index(drop=True)
spectralDf = add_fdr_to_csodiaq_output(overallDf)
peptideDf = add_fdr_to_csodiaq_output(overallDf, filterType='peptide')
spectralDf.to_csv(specFile, index=False)
peptideDf.to_csv(pepFile, index=False)
if protFile:
peptideProteinConnections = format_peptide_protein_connections(
peptideDf)
verifiedProteinDict = idp.find_valid_proteins(
peptideProteinConnections)
proteinDf = add_leading_protein_column(peptideDf, verifiedProteinDict)
tempProtDf = add_fdr_to_csodiaq_output(proteinDf,
filterType='leadingProtein')
proteinMetaInfoDict = tempProtDf.set_index(
'leadingProtein').T.to_dict()
proteinDf = remove_invalid_peptides_and_add_metadata(
proteinDf, proteinMetaInfoDict)
proteinDf = mark_peptides_unique_to_proteins(proteinDf)
proteinDf.to_csv(protFile, index=False)
def mgf_library_upload(fileName):
libMGF = mgf.read(fileName)
smf.print_milestone('Enter library dictionary upload: ')
lib = {}
id = 0
for spec in libMGF:
id += 1
key = (spec['params']['pepmass'][0], spec['params']['seq'])
charge = int(re.sub('[+-]', '', str(spec['params']['charge'][0])))
name = spec['params']['title']
if 'protein' in spec['params']: protein = spec['params']['protein']
else: protein = ''
if 'DECOY' in name: decoy = 1
else: decoy = 0
mz = spec['m/z array']
intensity = spec['intensity array']
intensity = [x**0.5 for x in intensity]
keyList = [id for x in mz]
peaks = list(tuple(zip(mz, intensity, keyList)))
peaks.sort(key=lambda x: x[1], reverse=True)
if len(peaks) > 10: peaks = peaks[:10]
peaks.sort(key=lambda x: x[0])
tempDict = {
'PrecursorCharge': charge,
'transition_group_id': name,
'ProteinName': protein,
'Peaks': peaks,
'ID': id,
'Decoy': decoy,
}
lib[key] = tempDict
return lib
def filter_fdr_output_for_targeted_reanalysis(fdrFile, proteins, heavy):
smf.print_milestone('Generate DISPA Targeted Reanalysis Files:')
fdrDf = pd.read_csv(fdrFile)
fdrDf = fdrDf[~fdrDf['protein'].str.contains('DECOY', na=False
)].reset_index(drop=True)
if heavy:
fdrDf = fdrDf[fdrDf['peptide'].str.endswith('R')
| fdrDf['peptide'].str.endswith('K')].reset_index(
drop=True)
if proteins:
fdrDf = fdrDf[fdrDf['uniquePeptide'] == 1].sort_values(
'ionCount', ascending=False).reset_index(drop=True)
fdrDf = fdrDf.groupby(['leadingProtein']).head(proteins).reset_index()
return fdrDf
def connect_mzxml_to_csodiaq_and_library(idFile, libFile, mzxmlFiles,
maxPeaks):
smf.print_milestone('Preparing Quantification Dictionaries:')
metadataToScanDict = create_mzxml_to_csodiaq_dict(mzxmlFiles[0])
fileType = idFile.split('.')[-1]
if fileType == 'csv': fragDf = pd.read_csv(idFile)
else: fragDf = pd.read_csv(idFile, sep='\t')
scanToCsodiaqDict, libMetadataToScanDict = connect_csodiaq_data_to_scans(
idFile, metadataToScanDict, fragDf)
scanToLibPeaksDict = pool_library_spectra_by_scan(libFile,
libMetadataToScanDict,
fragDf, maxPeaks)
return scanToCsodiaqDict, scanToLibPeaksDict
def heavy_light_quantification(fragDict, libDict, mzxmlFiles, outDir, massTol,
minMatch, ratioType, correction, hist):
finalDf = initialize_quantification_output(fragDict, libDict)
def initialize_ratio_dict_values():
return np.nan
for f in mzxmlFiles:
ppmDiffs = []
allSpectraMatch = QuantificationSpectraMatcher.QuantificationSpectraMatcher(
)
scanToNoiseIntensityCutoffDict = dict()
with mzxml.read(f, use_index=True) as file:
for scan in sorted(libDict.keys()):
spec = file.get_by_id(scan)
scanToNoiseIntensityCutoffDict[int(scan)] = np.mean(
sorted(spec['intensity array'])[:10]) / 2
expSpectrum = smf.format_spectra_for_pooling(spec,
scan,
sqrt=False)
expSpectrum.sort()
libSpectra = sorted(libDict[scan])
quantSpectraMatch = QuantificationSpectraMatcher.QuantificationSpectraMatcher(
)
quantSpectraMatch.compare_spectra(libSpectra, expSpectrum,
massTol, minMatch)
allSpectraMatch.extend_all_spectra(quantSpectraMatch)
if correction != -1:
allSpectraMatch.filter_by_corrected_ppm_window(
correction, hist, minMatch)
ratioDict = defaultdict(initialize_ratio_dict_values)
if len(allSpectraMatch.libraryIntensities) != 0:
ratioDict = allSpectraMatch.determine_ratios(
ratioDict, scanToNoiseIntensityCutoffDict, ratioType, minMatch)
finalDf[f] = [
ratioDict[(int(row['scan']), row['peptide'])]
for index, row in finalDf.iterrows()
]
smf.print_milestone('Finish SILAC Quantification')
return finalDf
def perform_spectra_pooling_and_analysis(querySpectraFile, outFile, lib,
tolerance, maxQuerySpectraToPool,
corrected, histFile):
smf.print_milestone('Begin Grouping Scans by m/z Windows:')
queWindowDict, queScanValuesDict = pool_scans_by_mz_windows(
querySpectraFile)
print('Number of Unpooled MS/MS Query Spectra: ' +
str(len(queScanValuesDict)))
print('Number of Pooled MS/MS Query Spectra/Mz Windows: ' +
str(len(queWindowDict)),
flush=True)
# To enhance the print experience, status prints will be given at intervals tailored to the number of identified windows.
# example: if there are 1-99 pooled query spectra, print statements are made after every pooled query spectra analysis is complete.
# if there are 100-999, print after every 10 pooled spectra. And so on.
printFriendlyCounter = 100
while printFriendlyCounter < len(queWindowDict):
printFriendlyCounter *= 10
printFriendlyCounter /= 100
allLibKeys, libIdToKeyDict, libIdToDecoyDict = gather_library_metadata(lib)
allSpectraMatches = IdentificationSpectraMatcher.IdentificationSpectraMatcher(
)
numWindowsAnalyzed = 0
prevtime = timer()
smf.print_milestone('Begin Pooled Spectra Analysis:')
with mzxml.read(querySpectraFile, use_index=True) as spectra:
for precMz_win, scans in queWindowDict.items():
top_mz = precMz_win[0] + precMz_win[1] / 2
bottom_mz = precMz_win[0] - precMz_win[1] / 2
libKeys = identify_lib_spectra_in_window(top_mz, bottom_mz,
allLibKeys)
if len(libKeys) == 0: continue
pooledLibSpectra = pool_lib_spectra(lib, libKeys)
pooledQueSpectra = []
for i in range(len(scans)):
scanNumber = scans[i]
queSpectrum = spectra.get_by_id(scanNumber)
pooledQueSpectra += smf.format_spectra_for_pooling(
queSpectrum, scanNumber)
if (i % maxQuerySpectraToPool == 0
and i != 0) or i == len(scans) - 1:
pooledQueSpectra.sort()
windowSpectraMatches = IdentificationSpectraMatcher.IdentificationSpectraMatcher(
)
windowSpectraMatches.compare_spectra(
pooledLibSpectra, pooledQueSpectra, tolerance,
libIdToDecoyDict)
allSpectraMatches.extend_all_spectra(windowSpectraMatches)
pooledQueSpectra.clear()
numWindowsAnalyzed += 1
if numWindowsAnalyzed % printFriendlyCounter == 0:
time = timer()
print('\nNumber of Pooled Experimental Spectra Analyzed: ' +
str(numWindowsAnalyzed))
print('Number of Spectra in Current Pooled Spectra: ' +
str(len(scans)))
print('Time Since Last Checkpoint: ' +
str(round(time - prevtime, 2)) + ' Seconds',
flush=True)
prevtime = time
smf.print_milestone('Begin FDR Analysis:')
maccCutoff = allSpectraMatches.find_score_fdr_cutoff()
if corrected != -1:
smf.print_milestone('Begin Correction Process:')
allSpectraMatches.filter_by_corrected_ppm_window(
corrected, maccCutoff, histFile)
smf.print_milestone('Begin Corrected FDR Analysis:')
maccCutoff = allSpectraMatches.find_score_fdr_cutoff()
smf.print_milestone('\nBegin Writing to File: ')
allSpectraMatches.write_output(outFile, querySpectraFile, maccCutoff,
queScanValuesDict, libIdToKeyDict, lib)
def traml_library_upload(fileName):
if fileName.endswith('.tsv'):
lib_df = pd.read_csv(fileName, sep='\t')
else:
lib_df = pd.read_csv(fileName)
smf.print_milestone('Enter library dictionary upload: ')
# Pan human and spectraST libraries have different column names. This normalizes the columns.
headings = traml_column_headings(lib_df.columns)
lib_df = lib_df.loc[:,
lib_df.columns.intersection([
headings['PrecursorMz'],
headings['FullUniModPeptideName'],
headings['PrecursorCharge'], headings['ProductMz'],
headings['LibraryIntensity'], headings[
'transition_group_id'], headings['ProteinName']
])]
lib_df = lib_df[[
headings['PrecursorMz'], headings['FullUniModPeptideName'],
headings['PrecursorCharge'], headings['ProductMz'],
headings['LibraryIntensity'], headings['transition_group_id'],
headings['ProteinName']
]]
lib_df.columns = [
'PrecursorMz', 'FullUniModPeptideName', 'PrecursorCharge', 'ProductMz',
'LibraryIntensity', 'transition_group_id', 'ProteinName'
]
lib_df['LibraryIntensity'] = [
x**0.5 for x in list(lib_df['LibraryIntensity'])
]
lib_df['ID'] = list(
zip(lib_df['PrecursorMz'].tolist(),
lib_df['FullUniModPeptideName'].tolist()))
mz_dict = lib_df.groupby("ID")['ProductMz'].apply(list).to_dict()
intensity_dict = lib_df.groupby("ID")['LibraryIntensity'].apply(
list).to_dict()
lib_df.drop_duplicates(subset="ID", inplace=True)
lib_df = lib_df.loc[:,
lib_df.columns.intersection([
'ID', 'PrecursorCharge', 'transition_group_id',
'ProteinName'
])]
lib_df.set_index("ID", drop=True, inplace=True)
lib = lib_df.to_dict(orient="index")
# pan human library formats are different, including how the peptides are matched to proteins (esp. decoys). This section of code adjusts for this discrepancy.
if headings['type'] == 'PanHuman':
for key, value in lib.items():
proteins = lib[key]['ProteinName'].split('/')
num = proteins.pop(0)
newProteins = [x for x in proteins if 'DECOY' not in x]
proteinStr = str(len(newProteins))
for x in newProteins:
if 'DECOY' in num: proteinStr += ('/DECOY_' + x)
else: proteinStr += ('/' + x)
lib[key]['ProteinName'] = proteinStr
id = 0
for key in lib:
id += 1
mz, intensity = (list(t) for t in zip(
*sorted(zip(mz_dict[key], intensity_dict[key]))))
keyList = [id for i in range(len(mz))]
peaks = list(tuple(zip(mz, intensity, keyList)))
peaks.sort(key=lambda x: x[1], reverse=True)
if len(peaks) > 10: peaks = peaks[:10]
peaks.sort(key=lambda x: x[0])
lib[key]['Peaks'] = peaks
lib[key]['ID'] = id
if 'DECOY' in lib[key]['ProteinName']: lib[key]['Decoy'] = 1
else: lib[key]['Decoy'] = 0
return lib