kmer_vectors.py

import numpy as np
import math
from collections import Counter
# note that this assumes that all base pairs are upper cased

def kmers(coll, k):
    n = len(coll)
    for i in range(0, n - k + 1):
        yield coll[i:i+k]

def encode(kmer):
    v = 0
    for i, c in enumerate(kmer):
        v += 4 ** i + {'A': 0, 'C': 1, 'T': 2, 'G' : 3}[c]
    return v

def mk_vec(coll, k):
    v = Counter()
    for kmer in kmers(coll, k):
        v[kmer] += 1
    return v

def euclidean_dist(lhs, rhs):
    ks = set(lhs.keys()) | set(rhs.keys())
    sm = 0
    for k in ks:
        delta = lhs[k] - rhs[k]
        sm += delta ** 2
    return math.sqrt(sm)

def manhattan_dist(lhs, rhs):
    ks = set(lhs.keys()) | set(rhs.keys())
    sm = 0
    for k in ks:
        delta = lhs[k] - rhs[k]
        sm += abs(delta)
    return sm