def test_overlapping_imperfect(self):
length = 22
overlap = 3
tiles = [t[2] for t in tile(protein_seq, length, overlap)]
assert len(tiles) == 2
assert all([len(t) == length for t in tiles])
assert tiles[0] == Seq("METMSDYSKEVSEALSALRGEL")
assert tiles[1] == Seq("GELSALSAAISNTVRAGSYSAP")
def test_overlapping_perfect(self):
length = 21
overlap = 3
tiles = [t[2] for t in tile(protein_seq, length, overlap)]
assert len(tiles) == 3
assert all([len(t) == length for t in tiles])
assert tiles[0] == Seq("METMSDYSKEVSEALSALRGE")
assert tiles[1] == Seq("RGELSALSAAISNTVRAGSYS")
assert tiles[2] == Seq("SYSAPVAKDCKAGHCDSKAVL")
def test_nonoverlapping_imperfect(self):
length = 25
overlap = 0
tiles = [t[2] for t in tile(protein_seq, length, overlap)]
assert len(tiles) == 2
assert all([len(t) == length for t in tiles])
assert tiles[0] == Seq("METMSDYSKEVSEALSALRGELSAL")
assert tiles[1] == Seq("SAAISNTVRAGSYSAPVAKDCKAGH")
assert sum(tiles, Seq("")) == protein_seq[:50]
def test_nonoverlapping_perfect(self):
length = 19
overlap = 0
tiles = [t[2] for t in tile(protein_seq, length, overlap)]
assert len(tiles) == 3
assert all([len(t) == length for t in tiles])
assert tiles[0] == Seq("METMSDYSKEVSEALSALR")
assert tiles[1] == Seq("GELSALSAAISNTVRAGSY")
assert tiles[2] == Seq("SAPVAKDCKAGHCDSKAVL")
assert sum(tiles, Seq("")) == protein_seq
def test_negative_length(self):
length = -10
overlap = 5
with raises(ValueError):
tiles = list(tile(protein_seq, length, overlap))
def test_overlap_longer_than_length(self):
length = 10
overlap = 15
with raises(ValueError):
tiles = list(tile(protein_seq, length, overlap))
def test_length_longer_than_seq(self):
length = len(protein_seq) + 5
overlap = 5
tiles = list(tile(protein_seq, length, overlap))
assert len(tiles) == 0
def test_short_protein(self):
length = 56
overlap = 2
assert len(list(tile(short_protein_seq, length, overlap))) == 0
overlap = 20
assert len(list(tile(short_protein_seq, length, overlap))) == 0
def test_zero_length(self):
length = 0
overlap = 5
with raises(ValueError):
tiles = list(tile(protein_seq, length, overlap))
