def similarity_score(kmers1, kmers2, scoring="jaccard"):
"""Compute Jaccard similarity index"""
# count of kmers in common
intersection = float(kmerizer.count_common(kmers1, kmers2))
if scoring == "jaccard":
# Use Jaccard similarity index
score = intersection / (kmers1.size + kmers2.size - intersection)
elif scoring == "minsize":
# Use intersection / min_size (proper subset scores 1.0)
score = intersection / min(kmers1.size, kmers2.size)
elif scoring == "meansize":
# Use mean size in denominator (used in Mash)
score = intersection / ((kmers1.size + kmers2.size) / 2)
elif scoring == "maxsize":
# Use intersection / max_size (proper subset scores min/max)
score = intersection / max(kmers1.size, kmers2.size)
elif scoring == "reference":
# Use intersection / size of reference (useful for comparing reads to
# assembled references)
score = intersection / kmers2.size
else:
assert scoring in (
"jaccard", "minsize", "maxsize", "meansize", "reference"), \
"unknown scoring method"
return score
def reference(kmers1, kmers2):
"""Assume k-merset 2 is the k-merset of a reference genome."""
intersection = kmerizer.count_common(kmers1, kmers2)
return intersection / kmers2.size
def subset(kmers1, kmers2):
"""Calculate the fraction of k-mers in k-merset 1 that are also in k-merset
2, useful to check whether k-merset 1 is a subset of another."""
intersection = kmerizer.count_common(kmers1, kmers2)
return intersection / kmers1.size
def maxsize(kmers1, kmers2):
intersection = kmerizer.count_common(kmers1, kmers2)
return intersection / max(kmers1.size, kmers2.size)
def meansize(kmers1, kmers2):
intersection = kmerizer.count_common(kmers1, kmers2)
return intersection / ((kmers1.size + kmers2.size) / 2)
def jaccard(kmers1, kmers2):
"""Computes jaccard similarity. Returns numerator and denominator
separately."""
intersection = kmerizer.count_common(kmers1, kmers2)
return intersection / (kmers1.size + kmers2.size - intersection)