/MichelaNGLo-human-protein-data

pickled files containing protein data for humans

MIT LicenseMIT

data

The whole dataset used by MichelaNGLo is large. For it you will have to contact Matteo or regenerate the data. This is only the human proteome.

See Deployment notes for installation.

Note that the protein parsing module is dependent on only the transpiler module, which is required for the structural analysis of a variant only —if unneeded alter analyse/__init__.py to not use PyMOL_structure_analyser.py.

Dictionaries

In this repo are some JSON files intended for the dictionary subfolder of the protein-data folder and a tsv for the reference subfolder. The rest are for decompression into the pickle/taxid9606 subfolder –taxon id 9606 is human.

Usage

The files are gip compressed pickle files labelled by Uniprot accession.

To read a file:

import os
from michelanglo_protein import ProteinCore
ProteinCore.settings.startup(data_folder='protein-data')
p = ProteinCore().gload(file=os.path.join('gpickle','Q86V25.gpz'))

from pprint import PrettyPrinter
pprint = PrettyPrinter().pprint
pprint(p.asdict())

Alternatively, importing ProteinAnalyser:

from michelanglo_protein import ProteinAnalyser
ProteinAnalyser.settings.startup(data_folder='protein-data')
p = ProteinAnalyser().gload(file=os.path.join('gpickle','Q86V25.gpz'))
p.mutation = Mutation('p.N127W')
if p.check_mutation():
    p.predict_effect()
    p.analyse_structure()
    pprint(p.asdict()['structural']
else:
    p.mutation_discrepancy()

Do note that the class attribute .settings is actually a singleton instance, so ProteinAnalyser.settings is the same as ProteinCore.settings. Whereas settings.startup() is compulsory, settings.retrieve_references(ask=False, refresh=False) is not and in the case of ProteinAnalyser only the csv of ELM classes will be needed within the data folder in references, but will complain only otherwise.