Added warning when using $ node symbol in CodeML trees.

Author: icwells

AlignmentProcessorReadMe.txt

@@ -243,9 +243,11 @@ in the bin/ directory.
 	but trimmed to ten characters (some programs still set a ten character
 	limit on the length of names, so AlignmentProcessor trims the names). 
 	The 07_CodeMLonDir.py script will save any nodes you have specified 
-	before sending a plain Newick tree to ape (which will not work if 
-	there are PAML node symbols). It will then add any nodes back into the
-	tree after it has been trimmed.
+	with a "#" before sending a plain Newick tree to ape (which will not 
+	work if there are PAML node symbols). It will then add any nodes back 
+	into the tree after it has been trimmed. AlignmentProcessor will not 
+	currently save nodes specified with "$" since it is difficult to 
+	determine where a nested clade begins and ends.
 
 # Invoking the Ka/Ks pipeline with a UCSC alignment:
 

bin/07_CodeMLonDir.py

@@ -10,7 +10,7 @@
 import re
 import os
 
-def readIn(path):
+def controlFiles(path):
     '''Reads input files and stores them in memory'''
     usertree = False
     with open(path + "codeml.ctl", "r") as control:
@@ -23,9 +23,8 @@ def readIn(path):
                 newline = "\ttreefile = " + path + "07_codeml/tmp.tree \n"
                 ctl.insert(i, newline)
             # Determine if a phylogenic tree is needed
-            elif "runmode" in line:
-                if "0" in line or "1" in line:
-                    usertree = True
+            if "runmode = 0" in line or "runmode = 1" in line:
+                usertree = True
     # Only run if a tree is required
     if usertree == True:
         with open(path + "codeml.tree", "r") as intree:
@@ -37,6 +36,11 @@ def readIn(path):
             i = i.replace("(", "")
             i = i.replace(")", "")
             i = i .replace(";", "")
+            if "$" in i:
+                print()
+                print("\tPlease only use the pound sign (#) to indicate nades.")
+                print()
+                quit()
             if "#" in i or "$" in i:
                 node = i.split()
                 nodes[node[0]] = node[1]
@@ -56,8 +60,12 @@ def readIn(path):
             reftree.write(tree)
     elif usertree == False:
         pass
+    return usertree, nodes, ctl
+
+def rmSeqs(path):
+    '''Creates a dictionary of sequences removed from specific genes'''
     with open(path + "internalStops.txt", "r") as stops:
-        # Creates a dictionary of sequences removed from specific genes
+        # Add sequences with internal stop codons
         rmseqs = {}
         for line in stops:
             splt = line.split("\t")
@@ -67,7 +75,17 @@ def readIn(path):
                 # Must typecast dictionary values in order to create a list of
                 # strings
                 rmseqs[splt[0]] = [str(splt[1]).rstrip(), ]
-    return usertree, nodes, ctl, rmseqs
+    with open(path + "Logs/04_CountBasesLog.txt", "r") as countlog:
+        # Add sequences which were removed due to low content to the dict.
+        for line in stops:
+            splt = line.split("\t")
+            if splt[0] in rmseqs:
+                rmseqs[splt[0]].append(str(splt[1]).rstrip())
+            else:
+                rmseqs[splt[0]] = [str(splt[1]).rstrip(), ]
+    with open(path + "Logs/04_CountBasesLog.txt", "r") as countlog:
+        
+    return rmseqs
 
 def runCodeML(usertree, path, ctl, rmseqs, nodes, retainStops):
     '''Runs CodeML on all files in a directory using temporary controlfiles.'''
@@ -79,7 +97,7 @@ def runCodeML(usertree, path, ctl, rmseqs, nodes, retainStops):
         geneid = filename.split(".")[0]
         outfile = path + "07_codeml/" + geneid
         tempctl = path + "07_codeml/tmp.ctl"
-        if retainStops == False and usertree == True:
+        if usertree == True:
             # Prune tree only if it will be used
             pruneTree(path, geneid, rmseqs, nodes)
         with open(tempctl, "w") as temp:
@@ -155,7 +173,10 @@ def main():
                 retainStops = True
         except IndexError:
             pass
-        usertree, nodes, ctl, rmseqs = readIn(path)
+        # Reads in required data
+        usertree, nodes, ctl = controlFiles(path)
+        rmseqs = rmSeqs(path)
+        # Calls CodeML for each input file
         runCodeML(usertree, path, ctl, rmseqs, nodes, retainStops)
 
 if __name__ == "__main__":