def scheme3(kmer):
'''
2 bit alphabet, BWT, MTF, shitty huffman encoding, 0 => 1, 1 => 01, 2 => 000, 3 => 001
Add 6 bits for eol position
'''
def huff_size(c):
return { 0: 1,
1: 2,
2: 3,
3: 3,
}[c]
trans_kmer, eol_idx = bwt.eol_format(bwt.bwt(kmer))
trans_kmer = mtf.mtf(trans_kmer)
size = 0
size += 6
i = 0
while i < len(trans_kmer):
size += huff_size(trans_kmer[i])
i += 1
return size
def test_mtf(self):
out = mtf.mtf("ACGATGTA")
self.assertEqual(out, [0,1,2,2,3,2,1,2])
def test_identity(self):
encoding = mtf.mtf("ACTGTGTA")
decoding = mtf.imtf(encoding)
self.assertEqual(''.join(decoding), "ACTGTGTA")
