crash fx

src/scripts/get_paths_fasta.py

@@ -4,14 +4,14 @@
 import os
 import graph_functions 
 from scaffolding import path_storage
+import networkx as nx
 
-
-def main():
-    rukki_path = sys.argv[1]
-    fasta_file = sys.argv[2]
-    graph = sys.argv[3]
-    out_fasta = sys.argv[4]
-    G = graph_functions.load_direct_graph(graph)
-    path_storage = path_storage.PathStorage(G)
-    path_storage.readFromFile(rukki_path)
-    path_storage.writeFasta(fasta_file, out_fasta)
+rukki_path = sys.argv[1]
+fasta_file = sys.argv[2]
+graph_file = sys.argv[3]
+out_fasta = sys.argv[4]
+G = nx.DiGraph()
+graph_functions.load_direct_graph(graph_file, G )
+path_storage = path_storage.PathStorage(G)
+path_storage.readFromFile(rukki_path)
+path_storage.writePathAsFasta(fasta_file, out_fasta)

src/scripts/graph_functions.py

@@ -233,3 +233,7 @@ def get_telomeric_nodes(telomere_locations_file, G):
                 aux_tel_nodes.add(telnode)
     return aux_tel_nodes, new_G
 
+
+def rc_seq(seq):
+    rc = {"A": "T", "T": "A", "G": "C", "C": "G", "N": "N"}
+    return "".join([rc[x] for x in seq[::-1]])

src/scripts/scaffolding/path_storage.py

@@ -103,7 +103,7 @@ def writePathAsFasta(self, input_fasta, output_fasta):
                 cur = line.strip()[1:]
             else:
                 seqs[cur] = line.strip()
-        with open(output_fasta, "a") as f:
+        with open(output_fasta, "w") as f:
             for id in sorted(self.paths.keys()):
                 f.write(f">{id}\n")
                 prev_over = 0
@@ -124,10 +124,10 @@ def writePathAsFasta(self, input_fasta, output_fasta):
                         if node[-1] == "+":
                             f.write(seqs[nor_node][prev_over:])
                         else:
-                            to_out = gf.rc_seq(seqs[nor_node][prev_over:])
+                            to_out = graph_functions.rc_seq(seqs[nor_node][prev_over:])
                             f.write(to_out[prev_over:])
                         if node_id < len(self.paths[id]) - 1 and self.paths[id][node_id + 1][:-1] in seqs:
-                            overlap = self.G.get_edge_data(nor_node, self.paths[id][node_id + 1][:-1])['overlap']
+                            overlap = self.G.get_edge_data(node, self.paths[id][node_id + 1])['overlap']
                         else:
                             overlap = 0
                 f.write(f"\n")