remove_singular_components.py

import copy
import sys
from Graph import Graph
import find_components
import functions

if len(sys.argv) != 3:
    print("You need to give the following arguments <input_gfa> <k>")
    sys.exit(0)

print("reading files")
new_graph = Graph()
new_graph.read_gfa(sys.argv[1])
print("deep copying")
new_graph.k = int(sys.argv[2])
new_graph1 = copy.deepcopy(new_graph)


# hacky stuff I need to remove later, just have it for testing
print("finding bubbles")
new_graph1.find_bubbles()
print("removing chains then finding components")
functions.remove_chains(new_graph1)
con_comp1 = find_components.connected_components(new_graph)
nodes_to_remove_from_original = []
new_graph1.nodes_reset()
for cc in con_comp1:
    if len(cc) == 1:
        nodes_to_remove_from_original.append(cc[0])


print(new_graph1)
print("removing singular CCs from original graph")
functions.remove_nodes(new_graph, nodes_to_remove_from_original)
new_graph.nodes_reset()
print("finding bubbles in original graph")
new_graph.find_bubbles()
print(new_graph)
print("Done finding bubbles")

longest = new_graph.max_chain()
longest_chain = new_graph.longest_chain()
print("Longest chain bubble-wise has {} bubbles and length"
      " of {}\n".format(longest[1], new_graph.bubble_chains[longest[0]].chain_length()))

print("Longest chain sequence-wise has {} bp and"
      " {} bubbles\n".format(longest_chain[1], new_graph.bubble_chains[longest_chain[0]].length()))

# print(new_graph.bubble_chains[longest].list_chain())
print("The bubble chains covered {}% nodes in the graph".format(
    (new_graph.nodes_in_chains() * 100) / len(new_graph.nodes))
)

total_seq = 0
for n in new_graph.nodes.values():
    total_seq += n.seq_len

print("The percentage of sequences covered by chains is {}%".format(
    ((new_graph.chains_coverage() - (new_graph.nodes_in_chains() * float(new_graph.k))) * 100) / (
            float(total_seq) - len(new_graph.nodes) * float(new_graph.k))
))