Python API

The components of pyAscore can be imported directly into python in order to build custom scripts for scoring the localization of algorithms. This allows advanced logic to be performed such as spectral preprocessing or combining scores for more confident localization. There are three main components that are described in the subsequent pages:

• File reading: Mass spectrometry data is usually disseminated in XML type files which can be read with the pyteomics python package. pyAscore provides some conveniece classes which build on this functionality and provide scan and PSM data in a convenient format for passing to scoring.

• Localization scoring: The main scoring functionality of pyAscore is available as a class so that users have the option to tailor scoring to individual scans.

• Looping over modified peptides: Internally, pyAscore relies on an iterator which generates the theoretical fragments for every permutation of modified sites. We provide access to this iterator to fascilitate the building of new localization algorithms in python.

Example script

The following is an example script which scores localization for PSMs in a single mass spectrometry run. It runs almost exactly the same pipeline as the command line version of pyAscore.

# 1) Read in PSMs
psm_file = "psms.pep.xml"
id_parser = IdentificationParser(psm_file, "pepXML")
psm_objects = id_parser.to_list()

# 2) Read in Spectra
spectra_file = "spectra.mzML"
spectra_parser = SpectraParser(spectra_file, "mzML")
spectra_objects = spectra_parser.to_dict()

# 3) Initialize pyAscore object
mod_mass = 79.966331
ascore = PyAscore(bin_size=100., n_top=10,
                  mod_group="STY",
                  mod_mass=mod_mass,
                  mz_error=.05)

# 4) Score PSMs
pyascore_results = []
for psm in psm_objects:
    # 4.1) Check for modification of interest
    mod_select = np.isclose(psm["mod_masses"], mod_mass)
    nmods = np.sum(mod_select)

    if nmods >= 1:
        # 4.2) Grab spectrum
        spectrum = spectra_objects[psm["scan"]]

        # 4.3) Gather other modifications into aux mods
        aux_mod_pos = psm["mod_positions"][~mod_select].astype(np.uint32)
        aux_mod_masses = psm["mod_masses"][~mod_select].astype(np.float32)

        # 4.4) Run scoring algorithm
        ascore.score(mz_arr = spectrum["mz_values"],
                     int_arr = spectrum["intensity_values"],
                     peptide = psm["peptide"],
                     n_of_mod = np.sum(mod_select),
                     max_fragment_charge = psm["charge_state"] - 1,
                     aux_mod_pos = aux_mod_pos,
                     aux_mod_mass = aux_mod_masses)

        # 4.5) Place scores into an object to use later
        pyascore_results.append({"scan" : psm["scan"],
                                 "localized_peptide" : ascore.best_sequence,
                                 "pepscore" : ascore.best_score,
                                 "ascores" : ";".join([str(s) for s in ascore.ascores])})

# 5) Make a dataframe of scores and write to a file
pyascore_results = pd.DataFrame.from_records(pyascore_results)
pyascore_results.to_csv("pyascore_results.tsv", sep="\t")