def reconstruction(kmers):
# Use De Bruijn approach to create adjacency dictionary from list of k-mers
adj_dict = debruijn2(kmers)
# Create dictionary that tracks remaining edges (those not yet taken)
remain_d = deepcopy(adj_dict)
# Initializes Counter that will track which nodes are unbalanced,
# which gives us the start and finish
find_ends = Counter()
# Go through each node, using the counter find_ends to count how many nodes
# are adjacent to the current node and subtract 1 for each of those adj
for row in remain_d.items():
dir_out = row[0]
dir_ins = row[1]
for ins in dir_ins:
find_ends[ins] -= 1
find_ends[dir_out] += 1
# The most_common function return a list of elements and their counts
# from the most common to the least.
# The starting node will be most common, as there is 1 less node
# adjacent to it, while the ending node will have 1 less adjacency
node = find_ends.most_common()[0][0]
# Begin to reconstruct the string
reconstructed = node
# Continue as long as there are edges remaining untaken
while len(remain_d) > 0:
node = remain_d.pop(node)[0]
reconstructed = reconstructed + node[-1]
return reconstructed
def reconstruction(kmers):
# Use De Bruijn approach to create adjacency dictionary from list of k-mers
adj_dict = debruijn2(kmers)
# Create dictionary that tracks remaining edges (those not yet taken)
remain_d = deepcopy(adj_dict)
# Initializes Counter that will track which nodes are unbalanced,
# which gives us the start and finish
find_ends = Counter()
# Go through each node, using the counter find_ends to count how many nodes
# are adjacent to the current node and subtract 1 for each of those adj
for row in remain_d.items():
dir_out = row[0]
dir_ins = row[1]
for ins in dir_ins:
find_ends[ins] -= 1
find_ends[dir_out] += 1
# The most_common function return a list of elements and their counts
# from the most common to the least.
# The starting node will be most common, as there is 1 less node
# adjacent to it, while the ending node will have 1 less adjacency
node = find_ends.most_common()[0][0]
# Begin to reconstruct the string
reconstructed = node
# Continue as long as there are edges remaining untaken
while len(remain_d) > 0:
node = remain_d.pop(node)[0]
reconstructed = reconstructed+node[-1]
return reconstructed
def contig(kmers):
# k = len(kmers[0])
adj_dict = debruijn2(kmers)
remain_d = deepcopy(adj_dict)
# Initialize empty lists in which nobranching paths and isolated cycles
# will be added
paths = []
cycles = []
# Initializes Counter that will track the number of OUTGOING edges
# for each node and another Counter to track INCOMING edges
v_out = Counter()
v_in = Counter()
# For each node, increment the OUT counter for each connecting node while
# also incrementing the IN counter for those connected nodes
for row in adj_dict.items():
dir_out = row[0]
dir_ins = row[1]
for ins in dir_ins:
v_in[ins] += 1
v_out[dir_out] += 1
# Now that we have the counts to determine if a node is one-in-one-out,
# we will iterate through the nodes to find nonbranching paths and remove
# them once they've been utilized
for row in v_out.items():
node = row[0]
count = row[1]
# If the node is NOT a 1-in-1-out node, it can serve as starting point
# because it marks a branch
if count > 1 or count != v_in[node]:
for i, edge in enumerate(adj_dict[node]):
path = node + edge[-1]
remain = adj_dict[node][i:]
remain_d[node] = remain
# If added edge is a a 1-in-1-out node, continue to extend the
# path until you hit a node that branches
while v_out[edge] == 1 and v_in[edge] == 1:
path = path + adj_dict[edge][0][-1]
# BUG: modifying edge inside a program which iterates
# through edge is not good practice - alternative approach?
edge = adj_dict[edge][0]
# Add path to collection of nonbranching paths
paths.append(path)
# Find isolated cycles in the graph
# ---QUESTION: if a cycle is isolated (disconnected from the rest of the graph)
# ---but has a branch in it, should it be added as a nonbranching path?
# ---Under this code, it is not.
# If the node is 1-in-1-out, it may be the start of an isolated cycle
elif count == 1 and v_in[node] == 1:
start = node
edge = adj_dict[node][0]
cycle = start + edge[-1]
# If added edge is a a 1-in-1-out node, continue to extend the
# path until you hit a node that branches (eliminating the path as
# a possible cycle) or the starting node (completing the cycle)
while v_out[edge] == 1 and v_in[edge] == 1:
cycle = cycle + adj_dict[edge][0][-1]
edge = adj_dict[edge][0]
if edge == start:
cycles.append(cycle)
break
# Filter cycles for uniques
uniq_cyc_nodes = []
for cyc in cycles:
# Check if the starting node of the cycle has already been used in a
# cycle. If it has not, add the cycle to path and its nodes to a tracker.
if cyc[0] not in uniq_cyc_nodes:
paths.append(cyc)
for i, n in enumerate(cyc):
uniq_cyc_nodes.append(n)
return paths
def universal(k):
#generate list of possible k-mers
kmers = condensed_list_b(k)
# Use De Bruijn approach to create adjacency dictionary from list of k-mers
adj_dict = debruijn2(kmers)
# Dictionary that tracks remaining edges (those not yet taken),
# initialized as the input adjacency dict
remain_d = deepcopy(adj_dict)
# Randomly select a starting node
node_i = random.choice(range(len(adj_dict)))
node = adj_dict.items()[node_i][0]
# Initialize string with the starting node
u_string = node
# print u_string
# Continue as long as there are edges that remain untaken
while len(remain_d) > 0:
# Check if any adjacencies remain for the node and if so, how many
value = remain_d.get(node)
# print "node is: "+node
# print "value: ",
# print value
# print remain_d
# If the node has no unused edges, we must be back at V0 (since the
# graph is balanced), but we also know there are remaining edges out
# there. We need to expand the circle until it encompasses all nodes.
if value == None:
# print "Value is none"
# To do so, iterate through the current "cycle" until we find a
# node with an unused edge. Make that the new V0.
for i, n in enumerate(u_string):
n = u_string[i:i+k-1]
if remain_d.get(n) > 0:
node = n
u_string = u_string[i:]+u_string[(k-1):i+(k-1)]
# print "new node is: "+str(node)
# print "u_string now is:"+u_string
break
# If the node has a single unused edge, simply use it to continue the
# cycle by adding it to the list 'cycle' and removing it from the dict
elif len(value) == 1:
# print "Value is one"
node = remain_d.pop(node)[0]
u_string = u_string+node[-1]
# print u_string
# If the node has multiple unused edges, randomly select one to add to
# the cycle and remove it out from the node's list of adjacencies.
elif len(value) > 1:
# print "Value is greater than one"
random_i = random.randrange(len(value))
pos_nodes = remain_d[node]
new_node = pos_nodes.pop(random_i)
remain_d[node] = pos_nodes
node = new_node
u_string = u_string+node[-1]
# print u_string
# Since the string is a circle, the last k-1 digits will be a repeat of the
# first k-1 digits and thus can be removed.
return u_string[:-(k-1)]