def test_pseudo_count(self):
# use example from
# http://biologie.univ-mrs.fr/upload/p202/01.4.PSSM_theory.pdf
alpha = unambiguous_dna
dna_align = MultipleSeqAlignment([
SeqRecord(Seq("AACCACGTTTAA", alpha), id="ID001"),
SeqRecord(Seq("CACCACGTGGGT", alpha), id="ID002"),
SeqRecord(Seq("CACCACGTTCGC", alpha), id="ID003"),
SeqRecord(Seq("GCGCACGTGGGG", alpha), id="ID004"),
SeqRecord(Seq("TCGCACGTTGTG", alpha), id="ID005"),
SeqRecord(Seq("TGGCACGTGTTT", alpha), id="ID006"),
SeqRecord(Seq("TGACACGTGGGA", alpha), id="ID007"),
SeqRecord(Seq("TTACACGTGCGC", alpha), id="ID008")
])
summary = SummaryInfo(dna_align)
expected = FreqTable({
"A": 0.325,
"G": 0.175,
"T": 0.325,
"C": 0.175
}, FREQ, unambiguous_dna)
ic = summary.information_content(e_freq_table=expected,
log_base=math.exp(1),
pseudo_count=1)
self.assertAlmostEqualList(summary.ic_vector, [
0.110, 0.090, 0.360, 1.290, 0.800, 1.290, 1.290, 0.80, 0.610,
0.390, 0.470, 0.040
],
places=2)
self.assertAlmostEqual(ic, 7.546, places=3)
def test_nucleotides(self):
filename = "GFF/multi.fna"
format = "fasta"
alignment = AlignIO.read(filename, format, alphabet=unambiguous_dna)
summary = SummaryInfo(alignment)
c = summary.dumb_consensus(ambiguous="N")
self.assertEqual(str(c), 'NNNNNNNN')
self.assertNotEqual(c.alphabet, unambiguous_dna)
self.assertTrue(isinstance(c.alphabet, DNAAlphabet))
c = summary.gap_consensus(ambiguous="N")
self.assertEqual(str(c), 'NNNNNNNN')
self.assertNotEqual(c.alphabet, unambiguous_dna)
self.assertTrue(isinstance(c.alphabet, DNAAlphabet))
expected = FreqTable({
"A": 0.25,
"G": 0.25,
"T": 0.25,
"C": 0.25
}, FREQ, unambiguous_dna)
m = summary.pos_specific_score_matrix(chars_to_ignore=['-'],
axis_seq=c)
self.assertEqual(
str(m), """ A C G T
N 2.0 0.0 1.0 0.0
N 1.0 1.0 1.0 0.0
N 1.0 0.0 2.0 0.0
N 0.0 1.0 1.0 1.0
N 1.0 2.0 0.0 0.0
N 0.0 2.0 1.0 0.0
N 1.0 2.0 0.0 0.0
N 0.0 2.0 1.0 0.0
""")
# Have a generic alphabet, without a declared gap char, so must tell
# provide the frequencies and chars to ignore explicitly.
ic = summary.information_content(e_freq_table=expected,
chars_to_ignore=['-'])
self.assertAlmostEqual(ic, 7.32029999423075, places=6)