def gen_read(strand, read_len):
startpos = random.randint(0, len(strand) - read_len - 1)
seq = list(strand[startpos : startpos + read_len])
for i in xrange(len(seq)):
probs = MISREAD_PROBS[MISREAD_BASE_TO_IDX[seq[i]]]
seq[i] = MISREAD_IDX_TO_BASE[utils.loaded_dice(probs)]
return "".join(seq)
def kpp(k, reads):
centroids = [prepare_centroid(random.choice(reads), reads)]
while len(centroids) < k:
sqrd_distances = [min_distance(centroids, read) ** 2 for read in reads]
sqrd_distances_sum = sum(sqrd_distances)
probs = [d / sqrd_distances_sum for d in sqrd_distances]
new_centroid_idx = utils.loaded_dice(probs)
new_centroid = prepare_centroid(reads[new_centroid_idx], reads)
if new_centroid in centroids:
continue # ensure that we don't have similar centroids
centroids.append(new_centroid)
return centroids
def prepare_centroid(centroid, reads):
if not any(c == "_" for c in centroid):
return centroid # centroid doesn't have missed bases, return original one
centroid_lst = list(centroid)
probs_nonnorm = [1. - distance(centroid, read) for read in reads]
for i in xrange(len(centroid)):
if centroid[i] != "_":
continue
# we should drop reads that contain "_" in this position
probs_nonnorm_copy = probs_nonnorm[:] # copying an array
for read_idx in xrange(len(reads)):
if reads[read_idx][i] == "_":
probs_nonnorm_copy[read_idx] = 0.0
if sum(probs_nonnorm_copy) == 0.0:
continue # all reads contain _ in this position, despair
donor_idx = utils.loaded_dice(normalize(probs_nonnorm_copy))
centroid_lst[i] = reads[donor_idx][i]
return "".join(centroid_lst)