laydi/test/workflows/affy_workflowtest.py

import unittest
import sys

sys.path.append("../..")
from workflows.affy_workflow import PhenotypeDataset

class PhenotypeDatasetTest(unittest.TestCase):
    def testEmptyData(self):
        # we have a list of cel-files, but no categories
        cel_data = """\
CEL
02-05-33
03-07-38
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals(['CEL', 'phenotypes'], dataset.get_dim_names())
        self.assertEquals(['02-05-33', '03-07-38'], dataset.get_identifiers('CEL'))
        self.assertEquals([], dataset.get_identifiers('phenotypes'))

    def testFloatData(self):
        cel_data = """\
CEL\tage
02-05-33\t8.5
03-07-38\t9
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals(['CEL', 'phenotypes'], dataset.get_dim_names())
        self.assertEquals(['age'], dataset.get_identifiers('phenotypes'))
        self.assertEquals([[8.5],
                           [9]], dataset.asarray().tolist())

    def testCategoryData(self):
        """Categories expand to one numeric column for each category choice."""
        cel_data = """\
CEL\tsick
02-05-33\tyes
03-07-38\tno
04-93-33\tyes
08-32-33\tmaybe
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals(['CEL', 'phenotypes'], dataset.get_dim_names())
        self.assertEquals(['sick-yes', 'sick-no', 'sick-maybe'], dataset.get_identifiers('phenotypes'))
        self.assertEquals([[1, 0, 0],
                           [0, 1, 0],
                           [1, 0, 0],
                           [0, 0, 1]], dataset.asarray().tolist())

    def testMultipleCategoriesAndFloats(self):
        cel_data = """\
CEL\tsex\tage\tinfected
02-05-33\tF\t8\tI
02-05-34\tF\t9\tN
02-05-35\tM\t8\tI
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals(['sex-F', 'sex-M', 'age', 'infected-I', 'infected-N'],
                          dataset.get_identifiers('phenotypes'))

        self.assertEquals([[1, 0, 8, 1, 0],
                           [1, 0, 9, 0, 1],
                           [0, 1, 8, 1, 0]], dataset.asarray().tolist())

    def testGetPhenotypeTable(self):
        cel_data = """\
CEL\tsex\tage\tinfected
02-05-33\tF\t8\tI
02-05-34\tF\t9\tN
02-05-35\tM\t8\tI
"""
        dataset = PhenotypeDataset(cel_data)

        self.assertEquals([['CEL', 'sex', 'age', 'infected'],
                           ['02-05-33', 'F', '8', 'I'],
                           ['02-05-34', 'F', '9', 'N'],
                           ['02-05-35', 'M', '8', 'I']], dataset.get_phenotype_table())

        # we can also get a sorted list
        new_order = ['02-05-35', '02-05-33', '02-05-34']
        self.assertEquals([['CEL', 'sex', 'age', 'infected'],
                           ['02-05-35', 'M', '8', 'I'],
                           ['02-05-33', 'F', '8', 'I'],
                           ['02-05-34', 'F', '9', 'N']], dataset.get_phenotype_table(new_order))

    def testGetCategories(self):
        cel_data = """\
CEL\tsex\tage\tinfected
02-05-33\tF\t8\tI
02-05-34\tF\t9\tN
02-05-35\tM\t8\tI
"""
        dataset = PhenotypeDataset(cel_data)

        self.assertEquals(set(['F', 'M', 'I', 'N']), set(dataset.get_categories()))

    def testGetFactors(self):
        cel_data = """\
CEL\tsex\tage\tinfected
02-05-33\tF\t8\tI
02-05-34\tF\t9\tN
02-05-35\tM\t8\tI
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals(set(["sex", "infected"]), dataset.get_factors(["F", "I"]))

    def testGetCategoryVariable(self):
        """Can get set/unset list for given category."""
        cel_data = """\
CEL\tsex\tage\tinfected
02-05-33\tF\t8\tI
02-05-34\tF\t9\tN
02-05-35\tM\t8\tI
"""
        dataset = PhenotypeDataset(cel_data)
        self.assertEquals([1, 1, 0], dataset.get_category_variable("F"))
        self.assertEquals([0, 0, 1], dataset.get_category_variable("M"))
        self.assertEquals([1, 0, 1], dataset.get_category_variable("I"))
        self.assertEquals([0, 1, 0], dataset.get_category_variable("N"))

if __name__=='__main__':
    unittest.main()