ActivityTableGenerator.py

import gzip


class ATGenerator:

    """

    Reads gzip files for activitydata and generates a bytearray with activitystatus.
    @author: Sebastian Beyvers
    @contact: sebastian.beyvers@med.uni-giessen.de

    """

    def __init__(self, repre):

        # Constructor with parameter representation: List of keywords with a corresponding index
        # Only the Index as byte will be appended to a bytearray.

        self.representation = repre

    def read_table(self, file):

        # Reads the activity-ensembl file ("/data/EnsemblData/release/organism/activity/*")
        # and returns the activity as bytearray, based on the representation in self.representation.
        # Only the index is saved -> see ActivityTable.py for current representation.

        activity_table = []
        with gzip.open(file, 'rb') as f:
            for line in f:
                for index, re in enumerate(self.representation):
                    if re in str(line):
                        activity_table.append(index)
                        break
        return bytearray(activity_table)