comparison match_library.py @ 5:b1d339e0147e

files in library reader

4:80075a4c6543

    '''
    Returns a Galaxy formatted list of tuples containing all possibilities for the
    GC-column types. Used by the library_lookup.xml tool
    @param library_file: given library file from which the list of GC-column types is extracted
    '''
    (data, header) = read_library(library_file)

    if 'columntype' not in header:
        raise IOError('Missing columns in ', library_file)

    # Filter data on column type
    column_type = header.index("columntype")
    amounts_in_list_dict = count_occurrence([row[column_type] for row in data])
    galaxy_output = [(str(a) + "(" + str(b) + ")", a, False) for a, b in amounts_in_list_dict.items()]
    return(galaxy_output)


def filter_column(library_file, column_type_name):
    '''
    Filters the Retention Index database on column type
    @param library_file: file containing the database
    @param column_type_name: column type to filter on
    '''
    (data, header) = read_library(library_file)

    if ('columntype' not in header or
        'columnphasetype' not in header):
        raise IOError('Missing columns in ', library_file)

    column_type = header.index("columntype")
    statphase = header.index("columnphasetype")

    # Filter data on colunn type name
    statphase_list = [line[statphase] for line in data if line[column_type] == column_type_name]
    amounts_in_list_dict = count_occurrence(statphase_list)
    galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
    return(sorted(galaxy_output))


def filter_column2(library_file, column_type_name, statphase):
    '''
    Filters the Retention Index database on column type
    @param library_file: file containing the database
    @param column_type_name: column type to filter on
    @param statphase: stationary phase of the column to filter on
    '''
    (data, header) = read_library(library_file)

    if ('columntype' not in header or
        'columnphasetype' not in header or
        'columnname' not in header):
        raise IOError('Missing columns in ', library_file)

    column_type_column = header.index("columntype")
    statphase_column = header.index("columnphasetype")
    column_name_column = header.index("columnname")

    # Filter data on given column type name and stationary phase
    statphase_list = [line[column_name_column] for line in data if line[column_type_column] == column_type_name and
                      line[statphase_column] == statphase]
    amounts_in_list_dict = count_occurrence(statphase_list)
    galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
    return(sorted(galaxy_output))


def read_library(filename):
    '''

    fill a Galaxy drop-down combo box.
    
    '''
    files = glob.glob(dir_name + "/*.txt")
    if len(files) == 0:
        raise Exception("Configuration error: no library files found in <galaxy-home-dir>/" + dir_name)
    else:
        galaxy_output = [(str(get_file_name_no_ext(file_name)), str(os.path.abspath(file_name)), False) for file_name in files]
    return(galaxy_output)
    
def get_file_name_no_ext(full_name):

5:b1d339e0147e

    '''
    Returns a Galaxy formatted list of tuples containing all possibilities for the
    GC-column types. Used by the library_lookup.xml tool
    @param library_file: given library file from which the list of GC-column types is extracted
    '''
    if library_file == "":
        galaxy_output = [("", "", False)]
    else:
        (data, header) = read_library(library_file)
    
        if 'columntype' not in header:
            raise IOError('Missing columns in ', library_file)
    
        # Filter data on column type
        column_type = header.index("columntype")
        amounts_in_list_dict = count_occurrence([row[column_type] for row in data])
        galaxy_output = [(str(a) + "(" + str(b) + ")", a, False) for a, b in amounts_in_list_dict.items()]
        
    return(galaxy_output)


def filter_column(library_file, column_type_name):
    '''
    Filters the Retention Index database on column type
    @param library_file: file containing the database
    @param column_type_name: column type to filter on
    '''
    if library_file == "":
        galaxy_output = [("", "", False)]
    else:
        (data, header) = read_library(library_file)
    
        if ('columntype' not in header or
            'columnphasetype' not in header):
            raise IOError('Missing columns in ', library_file)
    
        column_type = header.index("columntype")
        statphase = header.index("columnphasetype")
    
        # Filter data on colunn type name
        statphase_list = [line[statphase] for line in data if line[column_type] == column_type_name]
        amounts_in_list_dict = count_occurrence(statphase_list)
        galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
        
    return(sorted(galaxy_output))


def filter_column2(library_file, column_type_name, statphase):
    '''
    Filters the Retention Index database on column type
    @param library_file: file containing the database
    @param column_type_name: column type to filter on
    @param statphase: stationary phase of the column to filter on
    '''
    if library_file == "":
        galaxy_output = [("", "", False)]
    else:
        (data, header) = read_library(library_file)
    
        if ('columntype' not in header or
            'columnphasetype' not in header or
            'columnname' not in header):
            raise IOError('Missing columns in ', library_file)
    
        column_type_column = header.index("columntype")
        statphase_column = header.index("columnphasetype")
        column_name_column = header.index("columnname")
    
        # Filter data on given column type name and stationary phase
        statphase_list = [line[column_name_column] for line in data if line[column_type_column] == column_type_name and
                          line[statphase_column] == statphase]
        amounts_in_list_dict = count_occurrence(statphase_list)
        galaxy_output = [(str(a) + "(" + str(b) + ")", a, False)for a, b in amounts_in_list_dict.items()]
        
    return(sorted(galaxy_output))


def read_library(filename):
    '''

    fill a Galaxy drop-down combo box.
    
    '''
    files = glob.glob(dir_name + "/*.txt")
    if len(files) == 0:
        # Configuration error: no library files found in <galaxy-home-dir>/" + dir_name :
        galaxy_output = [("Configuration error: no library files found", "", False)]
    else:
        galaxy_output = [(str(get_file_name_no_ext(file_name)), str(os.path.abspath(file_name)), False) for file_name in files]
    return(galaxy_output)
    
def get_file_name_no_ext(full_name):