def read_phyloxml(input_file):
"""
Parses a pyhlogenetic tree in phyloxml-format.
:param str input_file: path to file
:return: ete2.Tree object
"""
from ete2 import Phyloxml
project = Phyloxml()
project.build_from_file(input_file)
return project.get_phylogeny()
def getRelevantEdges( adjGraph, t1, t2 ):
pxml = Phyloxml()
pxml.build_from_file(t1)
pxml.build_from_file(t2)
la = filter( lambda x : x.find('LOST') == -1, pxml.phylogeny[0].get_leaf_names() )
lb = filter( lambda x : x.find('LOST') == -1, pxml.phylogeny[1].get_leaf_names() )
#lb = filter( lambda x : x.find('LOST') == -1, map( getName, cogent.LoadTree(t2).tips() ) )
crossValidationEdges = filter( lambda (x,y) : ((x in la) and (y in lb)) or ((y in la) and (x in lb)) , adjGraph.edges() )
relevantEdges = filter( lambda (x,y) : ((x in la) or (x in lb)) and ((y in la) or (y in lb)) , adjGraph.edges() )
newGraph = nx.Graph()
newGraph.add_nodes_from( la + lb )
newGraph.add_edges_from( relevantEdges )
return newGraph, crossValidationEdges
def getTreeFromPhyloxml(xml, saveToFile="default.xml", delFile=True): """ Read a phylogeny tree from a phyloxml string and return a TreeClass object or a list of TreeClass object """ project = Phyloxml() fo=open(saveToFile, "w+") fo.write(xml) fo.close() project.build_from_file(saveToFile) treeList=[] for tree in project.get_phylogeny(): treeList.append(TreeClass.import_from_PhyloxmlTree(tree)) if(delFile): os.remove(saveToFile) if len(treeList)==1: return treeList[0] return treeList
def readScoreFile(fname, noself, randomize=False):
# The strings naming the proteins whose interaction was removed in
# this input
tstring = fname.split('@')[-2].split("#")
# Convert to upper case and make into an edge name
tedge = (tstring[0].upper(), tstring[1].upper())
# Read in the phylogenies for the orthology groups
treeDir = "../../Parana2Data/HerpesPPIs/trees/rearranged"#"dataOut_June17/rearranged"
baseFile = fname.split('/')[-1]
orthoGroup1 = baseFile.split('@')[0]
orthoGroup2 = baseFile.split('@')[1]
t1 = "{0}/{1}.xml.rooting.0.ntg.reconciled.0.ntg.rearrange.0.ntg".format(treeDir,orthoGroup1)
t2 = "{0}/{1}.xml.rooting.0.ntg.reconciled.0.ntg.rearrange.0.ntg".format(treeDir,orthoGroup2)
if ( not (os.path.exists(t1) and os.path.exists(t2)) ):
return None, None
# The extant (non ancestral, non lost) nodes from the two homology groups
getName = lambda x : x.Name.upper()
pxml = Phyloxml()
pxml.build_from_file(t1)
pxml.build_from_file(t2)
la = filter( lambda x : x.find('LOST') == -1, map( lambda x: x.upper(), pxml.phylogeny[0].get_leaf_names() ) )
lb = filter( lambda x : x.find('LOST') == -1, map( lambda x: x.upper(), pxml.phylogeny[1].get_leaf_names() ) )
# The set of all possible interactions among the two homology groups
#pe = list(itertools.product(la,la)) + list(itertools.product(la,lb)) + list(itertools.product(lb,lb))
possibleEndpoints = combinationsWithSelf(la) + list(itertools.product(la,lb)) + combinationsWithSelf(lb) \
if not (orthoGroup1 == orthoGroup2 ) else combinationsWithSelf(la)
# From among all possible endpoints, only those protein pairs that reside in the
# same species represent a potential edge
allPossibleEdges = filter( lambda (x,y): x.split('_')[-1] == y.split('_')[-1], possibleEndpoints )
# an edge has both endpoints in the set of extant nodes
inCurrentGroups = lambda e, x, y: (e[0] in x or e[0] in y) and (e[1] in x or e[1] in y)
# an edge is relevant if it's constrained to the current groups
relevantExtantEdges = [ e for e in ExtantNetwork.edges_iter() if inCurrentGroups(e,la,lb) ]
# the set of potential edges that don't appear in the input network
nonPresentEdgesMinusTarget = list(set([ x for x in allPossibleEdges if not ExtantNetwork.has_edge(x[0],x[1])]))
# the same as above but including our target edge
nonPresentEdges = nonPresentEdgesMinusTarget + [tedge]
import random
# Ancestral edges start with an N or R
ancestral = ['R','N']
# The node is valid if it is neither lost nor ancestral
isValidNode = lambda x : (x[0] not in ancestral) and (x.find('LOST') == -1)
# Is the edge u,v the target edge?
isCurrentEdge = lambda u,v : (u == tedge[0] and v == tedge[1]) or (u == tedge[1] and v == tedge[0])
isRealEdge = lambda u,v : (not isCurrentEdge(u,v)) and ExtantNetwork.has_edge(u,v)
isValidEdge = lambda u,v : ((isValidNode(u) and isValidNode(v)) and (not isRealEdge(u,v)))
# Is u,v one of the edges we wish to consider?
def inPotentialEdges(u,v) :
contains = (u,v) in nonPresentEdges or (v,u) in nonPresentEdges
if noself:
return u != v and contains
else:
return contains
def isEdge( se, p1, p2 ):
r = ((se.p1 == p1 and se.p2 == p2) or (se.p1 == p2 and se.p2 == p1))
return r
scoredEdges = []
nonEdgesWithProb = set( nonPresentEdges )
with open(fname,'rb') as ifile:
for l in ifile:
toks = l.rstrip().split()
p1 = toks[0].upper()
p2 = toks[1].upper()
s = float(toks[3])
if inPotentialEdges(p1,p2):
if randomize: s = random.uniform(0.0,1.0)
#if p1 == p2: s = 0.0
se = ScoredEdge(p1,p2,s)
scoredEdges.append( se )
nonEdgesWithProb.discard((p1,p2))
nonEdgesWithProb.discard((p2,p1))
rev = True
for u,v in (nonEdgesWithProb - set(nonPresentEdges)):
s = random.uniform(0.0,1.0) if randomize else 0.0
scoredEdges.append(ScoredEdge(u, v, s))
# cost = 0.0
# for u,v in nonPresentEdges:
# se = ScoredEdge(u,v,cost)
# fe = [ e for e in scoredEdges if isEdge(e, u, v) ]
# if len(fe) == 0:
# scoredEdges.append(se)
random.shuffle(scoredEdges)
scoredEdges = list(enumerate(sorted( scoredEdges, key=lambda x: x.score, reverse=rev )))
# print(len(scoredEdges))
# print(t1,t2)
# print("Target Edge = {0}".format(tedge))
# print("Extant Edges = {0}".format(relevantExtantEdges))
# print("Potential Edges = {0}".format(nonPresentEdges))
# print("Scored Edges = {0}".format(scoredEdges))
res = [ x for x in scoredEdges if isEdge(x[1], tedge[0], tedge[1]) ]
if len(res) > 0:
print(res)
# Prev (ISMB)
#print(res[0][0],float(len(nonPresentEdges)-1))
#return (res[0][0], float(len(nonPresentEdges)-1))
# New
#print(res[0][0],float(len(scoredEdges)-1))
return (res[0][0], float(len(scoredEdges)-1))
else:
raise 'Hell'
def main ():
global options, args
if options.verbose: print time.asctime(),
if options.verbose: print "load and parse newick file"
# TODO: read newick file
tree = Phylo.read(args[0],'newick')
# TODO: convert newick to phyloxml
treeXML = StringIO()
Phylo.write(tree,treeXML,'phyloxml')
# TODO: read phyloxml as ete object
hPhylotree = Phyloxml()
with tempinput(treeXML.getvalue()) as tempfilename:
hPhylotree.build_from_file(tempfilename)
# TODO: get the tree
tree2 = hPhylotree.get_phylogeny()[0]
if options.verbose: print time.asctime(),
if options.verbose: print "load and parse taxonomy file"
# TODO: read taxonomy file
tax = get_taxonomy(args[1])
# TODO: refine taxonomy annotation of internal node
tree2 = add_taxonomy_for_internal_branch(tree2,tax)
# TODO: refine tree node label
#tree2 = add_node_label(tree2,tax)
for node in tree2.traverse():
if not node.is_leaf():
label = "null"
for t in ['kingdom','phylum','class','order','family','genus','species']:
if len(tax[node.id][t])>3:
label = tax[node.id][t]
node.add_feature("mylabel",label)
# TODO: add node depth
depth={}
if options.depth:
with open(options.depth) as f:
for line in f:
(id,dep) = line.split()
depth[id] = float(dep)
# TODO: add color attribute
if options.depth:
for node in tree2.iter_leaves():
if depth[node.id] >= 10 and depth[node.id] < 100:
node.add_feature("color","#D8BFD8")
elif depth[node.id] >= 100 and depth[node.id] < 1000:
node.add_feature("color","#DDA0DD")
elif depth[node.id] >= 1000 and depth[node.id] < 5000:
node.add_feature("color","#EE82EE")
elif depth[node.id] >= 5000:
node.add_feature("color","#DA70D6")
else:
node.add_feature("color","#E6E6FA")
# TODO: set tree style
ts = TreeStyle()
ts.show_leaf_name = False
ts.layout_fn = tree_layout
# TODO: show tree2
#tree2.show(tree_style=ts)
tree2.render(args[0]+".png",dpi=2048,tree_style=ts)
import sys
import re
from StringIO import StringIO
from ete2 import Phyloxml, phyloxml
#Creates empty phyloxml document
project = Phyloxml()
# Loads newick tree
phylo = phyloxml.PhyloxmlTree(newick=sys.argv[1])
# Set basic tree info as a phyloxml phylogeny object
phylo.phyloxml_phylogeny.set_name("test_tree")
if len(phylo.children) <= 2:
phylo.phyloxml_phylogeny.set_rooted("true")
else:
phylo.phyloxml_phylogeny.set_rooted("false")
# Add the tree to the phyloxml project
project.add_phylogeny(phylo)
# Export phyloxml document
OUTPUT = StringIO()
project.export(OUTPUT)
# Some ad-hoc changes to the phyloxml formatted document to meet the schema definition
text = OUTPUT.getvalue()
text = text.replace("phy:", "")
text = re.sub('branch_length_attr="[^"]+"', "", text)
header = """
def main(argv):
input_file=''
title='Title'
label_internal_nodes = False
label_leaves = False
out_file=''
width=750
out_file_xml=''
plot_rectangular = False
common_kmer_data_path=''
taxonomic_names_on_leaves = False
try:
opts, args = getopt.getopt(argv,"h:i:lnrto:w:x:D:",["Help=","InputCommonKmerXFile=","LabelLeaves=", "LabelInternalNodes=","Rectangular=", "TaxonomicNamesOnLeaves=", "OutFile=","Width=","OutFileXML=","CommonKmerDataPath="])
except getopt.GetoptError:
print 'Unknown option, call using: ./PlotNJTree.py -i <InputCommonKmerXFile> -D <CommonKmerDataPath> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -r <RectangularPlotFlag> -t <TaxonomicNamesOnLeavesFlag> -o <OutFile.png> -x <Outfile.xml> -w <Width>'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print './PlotNJTree.py -i <InputCommonKmerXFile> -D <CommonKmerDataPath> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -r <RectangularPlotFlag> -t <TaxonomicNamesOnLeavesFlag> -o <OutFile.png> -x <Outfile.xml> -w <Width>'
sys.exit(2)
elif opt in ("-i", "--InputCommonKmerXFile"):
input_file = arg
elif opt in ("-l", "--LabelLeaves"):
label_leaves = True
elif opt in ("-n","--LabelInternalNodes"):
label_internal_nodes = True
elif opt in ("-o", "--OutFile"):
out_file = arg
elif opt in ("-w", "--Width"):
width = int(arg)
elif opt in ("-x", "--OutFileXML"):
out_file_xml = arg
elif opt in ("-D", "--CommonKmerDataPath"):
common_kmer_data_path = arg
elif opt in ("-r", "--Rectangular"):
plot_rectangular = True
elif opt in ("-t", "--TaxonomicNamesOnLeaves"):
taxonomic_names_on_leaves = True
#Read in the x vector
fid = open(input_file,'r')
x = map(lambda y: float(y),fid.readlines())
fid.close()
#Normalize the x vector
#x = map(lambda y: y/sum(x),x)
#Read in the taxonomy
taxonomy = list()
fid = open(os.path.join(common_kmer_data_path,"Taxonomy.txt"),'r')
for line in fid:
taxonomy.append('_'.join(line.split()[0].split("_")[1:])) #Just take the first line of the taxonomy (erasing the taxID)
fid.close()
#Read in the basis for the ckm matrices
x_file_names = list()
fid = open(os.path.join(common_kmer_data_path,"FileNames.txt"),'r')
for line in fid:
x_file_names.append(os.path.basename(line.strip()))
fid.close()
#Read in the common kmer matrix
f=h5py.File(os.path.join(common_kmer_data_path,'CommonKmerMatrix-30mers.h5'),'r')
ckm30=np.array(f['common_kmers'],dtype=np.float64)
f.close()
f=h5py.File(os.path.join(common_kmer_data_path,'CommonKmerMatrix-50mers.h5'),'r')
ckm50=np.array(f['common_kmers'],dtype=np.float64)
f.close()
ckm30_norm = np.multiply(ckm30,1/np.diag(ckm30))
ckm50_norm = np.multiply(ckm50,1/np.diag(ckm50))
num_rows = ckm30_norm.shape[0]
num_cols = ckm30_norm.shape[1]
names = x_file_names
matrix=list()
for i in range(num_rows):
matrix.append([.5*(1-.5*ckm30_norm[i,j]-.5*ckm30_norm[j,i])+.5*(1-.5*ckm50_norm[i,j]-.5*ckm50_norm[j,i]) for j in range(i+1)])
#Construct the tree. Note I could use RapidNJ here, but a few tests have shown that the trees that RapidNJ creates are rubbish.
dm = _DistanceMatrix(names, matrix)
constructor = DistanceTreeConstructor()
tree = constructor.nj(dm)
t=Tree(tree.format('newick'),format=1)
#tree.format('newick')
#Phylo.draw_ascii(tree)
#Now I will put internal nodes in a certain phylogenetic distance between the root and a given node.
#Function to insert a node at a given distance
def insert_node(t, name_to_insert, insert_above, dist_along):
insert_at_node = t.search_nodes(name=insert_above)[0]
parent = (t&insert_above).up
orig_branch_length = t.get_distance(insert_at_node,parent)
if orig_branch_length < dist_along:
raise ValueError("error: dist_along larger than orig_branch_length")
removed_node = insert_at_node.detach()
removed_node.dist = orig_branch_length - dist_along
added_node = parent.add_child(name=name_to_insert, dist=dist_along)
added_node.add_child(removed_node)
#Function to insert a node some % along a branch
def insert_hyp_node(t, leaf_name, percent):
total_dist = t.get_distance(t.name,leaf_name)
percent_dist = percent*total_dist
child_node = (t&leaf_name)
ancestor_node = (t&child_node.name).up
while t.get_distance(t.name, ancestor_node) > percent_dist:
child_node = ancestor_node
ancestor_node = (t&child_node.name).up
insert_node(t, leaf_name+"_"+str(percent), child_node.name, percent_dist-t.get_distance(t.name, ancestor_node))
#Insert hypothetical nodes
hyp_node_names = dict()
cutoffs = [.9,.8,.7,.6,.5,.4,.3,.2,.1]
cutoffs = map(lambda y: y**1.5,cutoffs)
for i in range(len(x_file_names)):
xi = x[i:len(x):len(x_file_names)]
for j in range(1,len(cutoffs)+1):
if xi[j]>0:
insert_hyp_node(t, x_file_names[i], cutoffs[j-1])
hyp_node_names[x_file_names[i]+"_"+str(cutoffs[j-1])] = [x_file_names[i], cutoffs[j-1], j-1] #in case there are "_" in the file names
#insert_hyp_node(t, x_file_names[i],.5/t.get_distance(t.name,t&x_file_names[i])*cutoffs[j])
#Now put the bubbles on the nodes
def layout(node):
#print(node)
if node.is_leaf():
if node.name in x_file_names:
#make reconstructed bubble
size = x[x_file_names.index(node.name)]
F = CircleFace(radius=500*math.sqrt(size), color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F,node, 0, position="branch-right")
if taxonomic_names_on_leaves:
nameFace = AttrFace("name", fsize=25, fgcolor='black',text_suffix="_"+taxonomy[x_file_names.index(node.name)])
faces.add_face_to_node(nameFace, node, 0, position="branch-right")
else:
nameFace = AttrFace("name", fsize=25, fgcolor='black')
faces.add_face_to_node(nameFace, node, 0, position="branch-right")
elif node.name in hyp_node_names: #Otherwise it's a hypothetical node, just use recon x
node_base_name = hyp_node_names[node.name][0]
percent = hyp_node_names[node.name][1]
if node_base_name in x_file_names:
idx = hyp_node_names[node.name][2]
size = x[x_file_names.index(node_base_name)+(idx+1)*len(x_file_names)]
F = CircleFace(radius=500*math.sqrt(size), color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F,node, 0, position="branch-right")
#print node
#print size
else:
size=0
else:
size=0
#print(size)
ts = TreeStyle()
ts.layout_fn = layout
if plot_rectangular:
ts.mode = "r"
else:
ts.mode = "c"
ts.show_leaf_name = False
ts.min_leaf_separation = 50
#Export the tree to a png image
t.render(out_file, w=width, units="mm", tree_style=ts)
#Export the xml file
project = Phyloxml()
phylo = phyloxml.PhyloxmlTree(newick=t.write(format=0, features=[]))
phylo.phyloxml_phylogeny.set_name(title)
project.add_phylogeny(phylo)
project.export(open(out_file_xml,'w'))
def readScoreFile(fname, noself, randomize=False):
# The strings naming the proteins whose interaction was removed in
# this input
tstring = fname.split('@')[-2].split("#")
# Convert to upper case and make into an edge name
tedge = (tstring[0].upper(), tstring[1].upper())
# Read in the phylogenies for the orthology groups
treeDir = "../../Parana2Data/HerpesPPIs/trees/rearranged" #"dataOut_June17/rearranged"
baseFile = fname.split('/')[-1]
orthoGroup1 = baseFile.split('@')[0]
orthoGroup2 = baseFile.split('@')[1]
t1 = "{0}/{1}.xml.rooting.0.ntg.reconciled.0.ntg.rearrange.0.ntg".format(
treeDir, orthoGroup1)
t2 = "{0}/{1}.xml.rooting.0.ntg.reconciled.0.ntg.rearrange.0.ntg".format(
treeDir, orthoGroup2)
if (not (os.path.exists(t1) and os.path.exists(t2))):
return None, None
# The extant (non ancestral, non lost) nodes from the two homology groups
getName = lambda x: x.Name.upper()
pxml = Phyloxml()
pxml.build_from_file(t1)
pxml.build_from_file(t2)
la = filter(lambda x: x.find('LOST') == -1,
map(lambda x: x.upper(), pxml.phylogeny[0].get_leaf_names()))
lb = filter(lambda x: x.find('LOST') == -1,
map(lambda x: x.upper(), pxml.phylogeny[1].get_leaf_names()))
# The set of all possible interactions among the two homology groups
#pe = list(itertools.product(la,la)) + list(itertools.product(la,lb)) + list(itertools.product(lb,lb))
possibleEndpoints = combinationsWithSelf(la) + list(itertools.product(la,lb)) + combinationsWithSelf(lb) \
if not (orthoGroup1 == orthoGroup2 ) else combinationsWithSelf(la)
# From among all possible endpoints, only those protein pairs that reside in the
# same species represent a potential edge
allPossibleEdges = filter(
lambda (x, y): x.split('_')[-1] == y.split('_')[-1], possibleEndpoints)
# an edge has both endpoints in the set of extant nodes
inCurrentGroups = lambda e, x, y: (e[0] in x or e[0] in y) and (e[
1] in x or e[1] in y)
# an edge is relevant if it's constrained to the current groups
relevantExtantEdges = [
e for e in ExtantNetwork.edges_iter() if inCurrentGroups(e, la, lb)
]
# the set of potential edges that don't appear in the input network
nonPresentEdgesMinusTarget = list(
set([
x for x in allPossibleEdges
if not ExtantNetwork.has_edge(x[0], x[1])
]))
# the same as above but including our target edge
nonPresentEdges = nonPresentEdgesMinusTarget + [tedge]
import random
# Ancestral edges start with an N or R
ancestral = ['R', 'N']
# The node is valid if it is neither lost nor ancestral
isValidNode = lambda x: (x[0] not in ancestral) and (x.find('LOST') == -1)
# Is the edge u,v the target edge?
isCurrentEdge = lambda u, v: (u == tedge[0] and v == tedge[1]) or (
u == tedge[1] and v == tedge[0])
isRealEdge = lambda u, v: (not isCurrentEdge(u, v)
) and ExtantNetwork.has_edge(u, v)
isValidEdge = lambda u, v: (
(isValidNode(u) and isValidNode(v)) and (not isRealEdge(u, v)))
# Is u,v one of the edges we wish to consider?
def inPotentialEdges(u, v):
contains = (u, v) in nonPresentEdges or (v, u) in nonPresentEdges
if noself:
return u != v and contains
else:
return contains
def isEdge(se, p1, p2):
r = ((se.p1 == p1 and se.p2 == p2) or (se.p1 == p2 and se.p2 == p1))
return r
scoredEdges = []
nonEdgesWithProb = set(nonPresentEdges)
with open(fname, 'rb') as ifile:
for l in ifile:
toks = l.rstrip().split()
p1 = toks[0].upper()
p2 = toks[1].upper()
s = float(toks[3])
if inPotentialEdges(p1, p2):
if randomize: s = random.uniform(0.0, 1.0)
#if p1 == p2: s = 0.0
se = ScoredEdge(p1, p2, s)
scoredEdges.append(se)
nonEdgesWithProb.discard((p1, p2))
nonEdgesWithProb.discard((p2, p1))
rev = True
for u, v in (nonEdgesWithProb - set(nonPresentEdges)):
s = random.uniform(0.0, 1.0) if randomize else 0.0
scoredEdges.append(ScoredEdge(u, v, s))
# cost = 0.0
# for u,v in nonPresentEdges:
# se = ScoredEdge(u,v,cost)
# fe = [ e for e in scoredEdges if isEdge(e, u, v) ]
# if len(fe) == 0:
# scoredEdges.append(se)
random.shuffle(scoredEdges)
scoredEdges = list(
enumerate(sorted(scoredEdges, key=lambda x: x.score, reverse=rev)))
# print(len(scoredEdges))
# print(t1,t2)
# print("Target Edge = {0}".format(tedge))
# print("Extant Edges = {0}".format(relevantExtantEdges))
# print("Potential Edges = {0}".format(nonPresentEdges))
# print("Scored Edges = {0}".format(scoredEdges))
res = [x for x in scoredEdges if isEdge(x[1], tedge[0], tedge[1])]
if len(res) > 0:
print(res)
# Prev (ISMB)
#print(res[0][0],float(len(nonPresentEdges)-1))
#return (res[0][0], float(len(nonPresentEdges)-1))
# New
#print(res[0][0],float(len(scoredEdges)-1))
return (res[0][0], float(len(scoredEdges) - 1))
else:
raise 'Hell'
from ete2 import Phyloxml
project = Phyloxml()
project.build_from_file("apaf.xml")
# Each tree contains the same methods as a PhyloTree object
for tree in project.get_phylogeny():
print tree
# you can even use rendering options
tree.show()
# PhyloXML features are stored in the phyloxml_clade attribute
for node in tree:
print "Node name:", node.name
for seq in node.phyloxml_clade.get_sequence():
for domain in seq.domain_architecture.get_domain():
domain_data = [domain.valueOf_, domain.get_from(), domain.get_to()]
print " Domain:", '\t'.join(map(str, domain_data))
def MakePlot(x, org_names, ckm30, ckm50, outgroup, outfile, outfilexml, sum_x):
#Make sure names are unique
names = org_names
for name in names:
if names.count(name)>1:
temp_name = name
i=1
for dummy in range(0,names.count(name)-1): #Don't change the last one, just to make sure we don't conflict with the outgroup
names[names.index(temp_name)] = temp_name + "_" + str(i)
i = i +1
#Normalize the x vector
x = map(lambda y: y/sum(x),x)
ckm30_norm = np.multiply(ckm30,1/np.diag(ckm30))
ckm50_norm = np.multiply(ckm50,1/np.diag(ckm50))
num_rows = ckm30_norm.shape[0]
num_cols = ckm30_norm.shape[1]
matrix=list()
for i in range(num_rows):
matrix.append([.5*(1-.5*ckm30_norm[i,j]-.5*ckm30_norm[j,i])+.5*(1-.5*ckm50_norm[i,j]-.5*ckm50_norm[j,i]) for j in range(i+1)])
#Make the list of distances (ave of the two ckm matrices)
ckm_ave_train = .5*ckm30_norm+.5*ckm50_norm
ckm_ave_train_dist = dict()
for i in range(len(org_names)):
ckm_ave_train_dist[org_names[i]] = [.5*ckm_ave_train[i,j]+.5*ckm_ave_train[j,i] for j in range(len(org_names))]
#Construct the tree. Note I could use RapidNJ here, but a few tests have shown that the trees that RapidNJ creates are rubbish.
dm = _DistanceMatrix(names, matrix)
constructor = DistanceTreeConstructor()
tree = constructor.nj(dm)
t=Tree(tree.format('newick'),format=1)
#tree.format('newick')
#Phylo.draw_ascii(tree)
#Now I will put internal nodes in a certain phylogenetic distance between the root and a given node.
#Function to insert a node at a given distance
def insert_node(t, name_to_insert, insert_above, dist_along):
insert_at_node = t.search_nodes(name=insert_above)[0]
parent = (t&insert_above).up
orig_branch_length = t.get_distance(insert_at_node,parent)
if orig_branch_length < dist_along:
raise ValueError("error: dist_along larger than orig_branch_length in PlotPackage.py")
removed_node = insert_at_node.detach()
removed_node.dist = orig_branch_length - dist_along
added_node = parent.add_child(name=name_to_insert, dist=dist_along)
added_node.add_child(removed_node)
#Function to insert a node some % along a branch, taking into account the ckm distances and nodes already created in the NJ tree (and what distance their descendants are from everyone else)
def insert_hyp_node(t, leaf_name, percent, ckm_ave_train_dist, org_names):
dists = map(lambda y: abs(y-percent), ckm_ave_train_dist[leaf_name])
nearby_indicies = list()
#Add all the organisms that are within 0.05 of the given percent
# for i in range(len(dists)):
# if dists[i]<=.05:
# nearby_indicies.append(i)
nearby_names = list()
#If there are no nearby indicies, add the closest organism to the given percent
if nearby_indicies==[]:
nearby_names.append(org_names[dists.index(min(dists))])
else:
for i in range(len(nearby_indicies)):
nearby_names.append(org_names[i])
mean_dist = np.mean(map(lambda y: ckm_ave_train_dist[leaf_name][org_names.index(y)],nearby_names))
nearby_names.append(leaf_name)
LCA = t.get_common_ancestor(nearby_names)
LCA_to_leaf_dist = t.get_distance(LCA,leaf_name)
#divide the dist to the right/left of the LCA node by the number of percentage points in there
if LCA.name==t.name:
percent_dist = percent*LCA_to_leaf_dist
if mean_dist <= percent:
child_node = (t&leaf_name)
else:
child_node = (t&nearby_names[0])#This means "go up from root" in the direction of the nearest guy
ancestor_node = (t&child_node.name).up
elif mean_dist <= percent:
percent_dist = t.get_distance(LCA) + abs(percent-mean_dist)*(LCA_to_leaf_dist)/(1-mean_dist)
child_node = (t&leaf_name)
ancestor_node = (t&child_node.name).up
else:
percent_dist = t.get_distance(LCA) - abs(percent-mean_dist)*(t.get_distance(LCA))/(mean_dist)
child_node = (t&leaf_name)
ancestor_node = (t&child_node.name).up
while t.get_distance(t.name, ancestor_node) > percent_dist:
child_node = ancestor_node
ancestor_node = (t&child_node.name).up
insert_node(t, leaf_name+"_"+str(percent), child_node.name, percent_dist-t.get_distance(t.name, ancestor_node))
#Set outgroup
if outgroup in names:
t.set_outgroup(t&outgroup) #I will need to check that this outgroup is actually one of the names...
else:
print("WARNING: the chosen outgroup " + outgroup + " is not in the given taxonomy: ")
print(names)
print("Proceeding without setting an outgroup. This may cause results to be uninterpretable.")
#Insert hypothetical nodes
hyp_node_names = dict()
cutoffs = [.9,.8,.7,.6,.5,.4,.3,.2,.1]
cutoffs = [-.5141*(val**3)+1.0932*(val**2)+0.3824*val for val in cutoffs]
for i in range(len(org_names)):
xi = x[i:len(x):len(org_names)]
for j in range(1,len(cutoffs)+1):
if xi[j]>0:
insert_hyp_node(t, org_names[i], cutoffs[j-1],ckm_ave_train_dist, org_names)
hyp_node_names[org_names[i]+"_"+str(cutoffs[j-1])] = [org_names[i], cutoffs[j-1], j-1] #in case there are "_" in the file names
size_factor=250
font_size=55
#Now put the bubbles on the nodes
def layout(node):
node_style = NodeStyle()
node_style["hz_line_width"] = 10
node_style["vt_line_width"] = 10
node.set_style(node_style)
#print(node)
if node.is_leaf():
if node.name in org_names:
#make reconstructed bubble
size = x[org_names.index(node.name)]
F = CircleFace(radius=size_factor*math.sqrt(size), color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F,node, 0, position="branch-right")
#Denote that this was a training organism
nameFace = AttrFace("name", fsize=font_size, fgcolor='black')
faces.add_face_to_node(nameFace, node, 0, position="branch-right")
elif node.name in hyp_node_names: #Otherwise it's a hypothetical node, just use recon x
node_base_name = hyp_node_names[node.name][0]
percent = hyp_node_names[node.name][1]
if node_base_name in org_names:
idx = hyp_node_names[node.name][2]
size = x[org_names.index(node_base_name)+(idx+1)*len(org_names)]
F = CircleFace(radius=size_factor*math.sqrt(size), color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F,node, 0, position="branch-right")
#This is if I want the names of the hypothetical nodes to be printed as well
#nameFace = AttrFace("name", fsize=font_size, fgcolor='black')
#faces.add_face_to_node(nameFace, node, 0, position="branch-right")
else:
size=0
else:
size=0
ts = TreeStyle()
ts.layout_fn = layout
ts.mode = "r"
#ts.mode = "c"
ts.scale = 2*1000
ts.show_leaf_name = False
ts.min_leaf_separation = 50
F = CircleFace(radius=.87*size_factor, color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
ts.legend.add_face(F,0)
ts.legend.add_face(TextFace(" Inferred relative abundance",fsize=1.5*font_size,fgcolor="Blue"),1)
ts.legend.add_face(TextFace(" Total absolute abundance depicted " + str(sum_x)[0:8], fsize=1.5*font_size,fgcolor="Black"),1)
ts.legend_position=4
#t.show(tree_style=ts)
t.render(outfile, w=550, units="mm", tree_style=ts)
#Redner the XML file
project = Phyloxml()
phylo = phyloxml.PhyloxmlTree(newick=t.write(format=0, features=[]))
project.add_phylogeny(phylo)
project.export(open(outfilexml,'w'))
from ete2 import Phyloxml
project = Phyloxml()
project.build_from_file("testTree.xml")
# Each tree contains the same methods as a PhyloTree object
for tree in project.get_phylogeny():
print tree
# you can even use rendering options
tree.show()
# PhyloXML features are stored in the phyloxml_clade attribute
for node in tree:
print "Node name:", node.name
for seq in node.phyloxml_clade.get_sequence():
for domain in seq.domain_architecture.get_domain():
domain_data = [domain.valueOf_, domain.get_from(), domain.get_to()]
print " Domain:", '\t'.join(map(str, domain_data))
from ete2 import Phyloxml, phyloxml
import random
project = Phyloxml()
# Creates a random tree
phylo = phyloxml.PhyloxmlTree()
phylo.populate(5, random_branches=True)
phylo.phyloxml_phylogeny.set_name("test_tree")
# Add the tree to the phyloxml project
project.add_phylogeny(phylo)
print project.get_phylogeny()[0]
# /-iajom
# /---|
# | \-wiszh
#----|
# | /-xrygw
# \---|
# | /-gjlwx
# \---|
# \-ijvnk
# Trees can be operated as normal ETE trees
phylo.show()
# Export the project as phyloXML format
project.export()
# <phy:Phyloxml xmlns:phy="http://www.phyloxml.org/1.10/phyloxml.xsd">
def MakePlot(x, org_names, ckm30, ckm50, outgroup, outfile, outfilexml, sum_x):
#Make sure names are unique
names = org_names
for name in names:
if names.count(name) > 1:
temp_name = name
i = 1
for dummy in range(
0,
names.count(name) - 1
): #Don't change the last one, just to make sure we don't conflict with the outgroup
names[names.index(temp_name)] = temp_name + "_" + str(i)
i = i + 1
#Normalize the x vector
x = map(lambda y: y / sum(x), x)
ckm30_norm = np.multiply(ckm30, 1 / np.diag(ckm30))
ckm50_norm = np.multiply(ckm50, 1 / np.diag(ckm50))
num_rows = ckm30_norm.shape[0]
num_cols = ckm30_norm.shape[1]
matrix = list()
for i in range(num_rows):
matrix.append([
.5 * (1 - .5 * ckm30_norm[i, j] - .5 * ckm30_norm[j, i]) + .5 *
(1 - .5 * ckm50_norm[i, j] - .5 * ckm50_norm[j, i])
for j in range(i + 1)
])
#Make the list of distances (ave of the two ckm matrices)
ckm_ave_train = .5 * ckm30_norm + .5 * ckm50_norm
ckm_ave_train_dist = dict()
for i in range(len(org_names)):
ckm_ave_train_dist[org_names[i]] = [
.5 * ckm_ave_train[i, j] + .5 * ckm_ave_train[j, i]
for j in range(len(org_names))
]
#Construct the tree. Note I could use RapidNJ here, but a few tests have shown that the trees that RapidNJ creates are rubbish.
dm = _DistanceMatrix(names, matrix)
constructor = DistanceTreeConstructor()
tree = constructor.nj(dm)
t = Tree(tree.format('newick'), format=1)
#tree.format('newick')
#Phylo.draw_ascii(tree)
#Now I will put internal nodes in a certain phylogenetic distance between the root and a given node.
#Function to insert a node at a given distance
def insert_node(t, name_to_insert, insert_above, dist_along):
insert_at_node = t.search_nodes(name=insert_above)[0]
parent = (t & insert_above).up
orig_branch_length = t.get_distance(insert_at_node, parent)
if orig_branch_length < dist_along:
raise ValueError(
"error: dist_along larger than orig_branch_length in PlotPackage.py"
)
removed_node = insert_at_node.detach()
removed_node.dist = orig_branch_length - dist_along
added_node = parent.add_child(name=name_to_insert, dist=dist_along)
added_node.add_child(removed_node)
#Function to insert a node some % along a branch, taking into account the ckm distances and nodes already created in the NJ tree (and what distance their descendants are from everyone else)
def insert_hyp_node(t, leaf_name, percent, ckm_ave_train_dist, org_names):
dists = map(lambda y: abs(y - percent), ckm_ave_train_dist[leaf_name])
nearby_indicies = list()
#Add all the organisms that are within 0.05 of the given percent
# for i in range(len(dists)):
# if dists[i]<=.05:
# nearby_indicies.append(i)
nearby_names = list()
#If there are no nearby indicies, add the closest organism to the given percent
if nearby_indicies == []:
nearby_names.append(org_names[dists.index(min(dists))])
else:
for i in range(len(nearby_indicies)):
nearby_names.append(org_names[i])
mean_dist = np.mean(
map(lambda y: ckm_ave_train_dist[leaf_name][org_names.index(y)],
nearby_names))
nearby_names.append(leaf_name)
LCA = t.get_common_ancestor(nearby_names)
LCA_to_leaf_dist = t.get_distance(LCA, leaf_name)
#divide the dist to the right/left of the LCA node by the number of percentage points in there
if LCA.name == t.name:
percent_dist = percent * LCA_to_leaf_dist
if mean_dist <= percent:
child_node = (t & leaf_name)
else:
child_node = (
t & nearby_names[0]
) #This means "go up from root" in the direction of the nearest guy
ancestor_node = (t & child_node.name).up
elif mean_dist <= percent:
percent_dist = t.get_distance(LCA) + abs(percent - mean_dist) * (
LCA_to_leaf_dist) / (1 - mean_dist)
child_node = (t & leaf_name)
ancestor_node = (t & child_node.name).up
else:
percent_dist = t.get_distance(LCA) - abs(percent - mean_dist) * (
t.get_distance(LCA)) / (mean_dist)
child_node = (t & leaf_name)
ancestor_node = (t & child_node.name).up
while t.get_distance(t.name, ancestor_node) > percent_dist:
child_node = ancestor_node
ancestor_node = (t & child_node.name).up
insert_node(t, leaf_name + "_" + str(percent), child_node.name,
percent_dist - t.get_distance(t.name, ancestor_node))
#Set outgroup
if outgroup in names:
t.set_outgroup(
t & outgroup
) #I will need to check that this outgroup is actually one of the names...
else:
print("WARNING: the chosen outgroup " + outgroup +
" is not in the given taxonomy: ")
print(names)
print(
"Proceeding without setting an outgroup. This may cause results to be uninterpretable."
)
#Insert hypothetical nodes
hyp_node_names = dict()
cutoffs = [.9, .8, .7, .6, .5, .4, .3, .2, .1]
cutoffs = [
-.5141 * (val**3) + 1.0932 * (val**2) + 0.3824 * val for val in cutoffs
]
for i in range(len(org_names)):
xi = x[i:len(x):len(org_names)]
for j in range(1, len(cutoffs) + 1):
if xi[j] > 0:
insert_hyp_node(t, org_names[i], cutoffs[j - 1],
ckm_ave_train_dist, org_names)
hyp_node_names[org_names[i] + "_" + str(cutoffs[j - 1])] = [
org_names[i], cutoffs[j - 1], j - 1
] #in case there are "_" in the file names
size_factor = 250
font_size = 55
#Now put the bubbles on the nodes
def layout(node):
node_style = NodeStyle()
node_style["hz_line_width"] = 10
node_style["vt_line_width"] = 10
node.set_style(node_style)
#print(node)
if node.is_leaf():
if node.name in org_names:
#make reconstructed bubble
size = x[org_names.index(node.name)]
F = CircleFace(radius=size_factor * math.sqrt(size),
color="RoyalBlue",
style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F, node, 0, position="branch-right")
#Denote that this was a training organism
nameFace = AttrFace("name", fsize=font_size, fgcolor='black')
faces.add_face_to_node(nameFace,
node,
0,
position="branch-right")
elif node.name in hyp_node_names: #Otherwise it's a hypothetical node, just use recon x
node_base_name = hyp_node_names[node.name][0]
percent = hyp_node_names[node.name][1]
if node_base_name in org_names:
idx = hyp_node_names[node.name][2]
size = x[org_names.index(node_base_name) +
(idx + 1) * len(org_names)]
F = CircleFace(radius=size_factor * math.sqrt(size),
color="RoyalBlue",
style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F, node, 0, position="branch-right")
#This is if I want the names of the hypothetical nodes to be printed as well
#nameFace = AttrFace("name", fsize=font_size, fgcolor='black')
#faces.add_face_to_node(nameFace, node, 0, position="branch-right")
else:
size = 0
else:
size = 0
ts = TreeStyle()
ts.layout_fn = layout
ts.mode = "r"
#ts.mode = "c"
ts.scale = 2 * 1000
ts.show_leaf_name = False
ts.min_leaf_separation = 50
F = CircleFace(radius=.87 * size_factor, color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
ts.legend.add_face(F, 0)
ts.legend.add_face(
TextFace(" Inferred relative abundance",
fsize=1.5 * font_size,
fgcolor="Blue"), 1)
ts.legend.add_face(
TextFace(" Total absolute abundance depicted " + str(sum_x)[0:8],
fsize=1.5 * font_size,
fgcolor="Black"), 1)
ts.legend_position = 4
#t.show(tree_style=ts)
t.render(outfile, w=550, units="mm", tree_style=ts)
#Redner the XML file
project = Phyloxml()
phylo = phyloxml.PhyloxmlTree(newick=t.write(format=0, features=[]))
project.add_phylogeny(phylo)
project.export(open(outfilexml, 'w'))
def main(argv):
input_file=''
title='Title'
label_internal_nodes = False
label_leaves = False
out_file=''
width=750
out_file_xml=''
try:
opts, args = getopt.getopt(argv,"h:i:t:lno:w:x:",["Help=","InputFile=","Title=","LabelLeaves=", "LabelInternalNodes=","OutFile=","Width=","OutFileXML="])
except getopt.GetoptError:
print 'Unknown option, call using: ./PlotTree.py -i <InputCAMIFile> -t <Title> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -o <OutFile.png> -x <Outfile.xml> -w <Width>'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print './PlotTree.py -i <InputCAMIFile> -t <Title> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -o <OutFile> -x <OutFile.xml> -w <Width>'
sys.exit(2)
elif opt in ("-i", "--InputFile"):
input_file = arg
elif opt in ("-t", "--Title"):
title = arg
elif opt in ("-l", "--LabelLeaves"):
label_leaves = True
elif opt in ("-n","--LabelInternalNodes"):
label_internal_nodes = True
elif opt in ("-o", "--OutFile"):
out_file = arg
elif opt in ("-w", "--Width"):
width = int(arg)
elif opt in ("-x", "--OutFileXML"):
out_file_xml = arg
schema_names = COLOR_SCHEMES.keys()
#Read the common kmer profile
ckm_tax_paths = []
ckm_name_to_perc = dict()
fid = open(input_file,'r')
file = fid.readlines()
fid.close()
#Put placeholders in for missing names like: "||" -> "|NA1|"
file_noblank = list()
i=0
for line in file:
while "||" in line:
line = line.replace("||","|NONAME|",1)
i = i+1
file_noblank.append(line)
#Get the names and weights
for line in file_noblank:
if line[0]!='#' and line[0]!='@' and line[0]!='\n': #Don't parse comments or blank lines
temp = line.split()[3] #Get the names
ckm_tax_paths.append(temp)
ckm_name_to_perc[temp.split("|")[-1]] = line.split()[-1] #Get the weights
#Create the tree
t=Tree()
names_to_nodes = dict()
for i in range(0,len(ckm_tax_paths)):
split_tax_path = ckm_tax_paths[i].split("|")
if len(split_tax_path)==1: #If len==1, then it's a superkingdom
names_to_nodes[split_tax_path[0]] = t.add_child(name=split_tax_path[0]) #connect directly to tree
else:
if split_tax_path[-2] in names_to_nodes: #If the parent is already in the tree, add to tree
names_to_nodes[split_tax_path[-1]] = names_to_nodes[split_tax_path[-2]].add_child(name=split_tax_path[-1])
else: #Otherwise iterate up until we have something that is in the tree
j=2
while split_tax_path[-j]=="NONAME":
j = j + 1
#This skips over the NONAMES
names_to_nodes[split_tax_path[-1]] = names_to_nodes[split_tax_path[-j]].add_child(name=split_tax_path[-1])
#Show the tree
#print t.get_ascii(show_internal=True)
#scheme = random.sample(schema_names, 1)[0] #'set2' is nice,
scheme = 'set2'
def layout(node):
if node.name in ckm_name_to_perc:
ckm_perc = float(ckm_name_to_perc[node.name])
else:
ckm_perc = 0
F = CircleFace(radius=3.14*math.sqrt(ckm_perc), color="RoyalBlue", style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F,node, 0, position="branch-right")
if label_internal_nodes:
faces.add_face_to_node(TextFace(node.name, fsize=7),node, 0, position="branch-top")
ts = TreeStyle()
ts.layout_fn = layout
ts.mode = "r"
ts.show_leaf_name = label_leaves
ts.min_leaf_separation = 50
ts.title.add_face(TextFace(title, fsize=20), column=0)
#Export the tree to a png image
t.render(out_file, w=width, units="mm", tree_style=ts)
#Export the xml file
project = Phyloxml()
phylo = phyloxml.PhyloxmlTree(newick=t.write(format=0, features=[]))
phylo.phyloxml_phylogeny.set_name(title)
project.add_phylogeny(phylo)
project.export(open(out_file_xml,'w'))
import sys
import re
from StringIO import StringIO
from ete2 import Phyloxml, phyloxml
#Creates empty phyloxml document
project = Phyloxml()
# Loads newick tree
phylo = phyloxml.PhyloxmlTree(newick=sys.argv[1])
# Set basic tree info as a phyloxml phylogeny object
phylo.phyloxml_phylogeny.set_name("test_tree")
if len(phylo.children) <= 2:
phylo.phyloxml_phylogeny.set_rooted("true")
else:
phylo.phyloxml_phylogeny.set_rooted("false")
# Add the tree to the phyloxml project
project.add_phylogeny(phylo)
# Export phyloxml document
OUTPUT = StringIO()
project.export(OUTPUT)
# Some ad-hoc changes to the phyloxml formatted document to meet the schema definition
text = OUTPUT.getvalue()
text = text.replace("phy:", "")
text = re.sub('branch_length_attr="[^"]+"', "", text)
header = """
def main(argv):
input_file = ''
title = 'Title'
label_internal_nodes = False
label_leaves = False
out_file = ''
width = 750
out_file_xml = ''
plot_rectangular = False
common_kmer_data_path = ''
taxonomic_names_on_leaves = False
try:
opts, args = getopt.getopt(argv, "h:i:lnrto:w:x:D:", [
"Help=", "InputCommonKmerXFile=", "LabelLeaves=",
"LabelInternalNodes=", "Rectangular=", "TaxonomicNamesOnLeaves=",
"OutFile=", "Width=", "OutFileXML=", "CommonKmerDataPath="
])
except getopt.GetoptError:
print 'Unknown option, call using: ./PlotNJTree.py -i <InputCommonKmerXFile> -D <CommonKmerDataPath> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -r <RectangularPlotFlag> -t <TaxonomicNamesOnLeavesFlag> -o <OutFile.png> -x <Outfile.xml> -w <Width>'
sys.exit(2)
for opt, arg in opts:
if opt == '-h':
print './PlotNJTree.py -i <InputCommonKmerXFile> -D <CommonKmerDataPath> -l <LabelLeavesFlag> -n <LabelInternalNodesFlag> -r <RectangularPlotFlag> -t <TaxonomicNamesOnLeavesFlag> -o <OutFile.png> -x <Outfile.xml> -w <Width>'
sys.exit(2)
elif opt in ("-i", "--InputCommonKmerXFile"):
input_file = arg
elif opt in ("-l", "--LabelLeaves"):
label_leaves = True
elif opt in ("-n", "--LabelInternalNodes"):
label_internal_nodes = True
elif opt in ("-o", "--OutFile"):
out_file = arg
elif opt in ("-w", "--Width"):
width = int(arg)
elif opt in ("-x", "--OutFileXML"):
out_file_xml = arg
elif opt in ("-D", "--CommonKmerDataPath"):
common_kmer_data_path = arg
elif opt in ("-r", "--Rectangular"):
plot_rectangular = True
elif opt in ("-t", "--TaxonomicNamesOnLeaves"):
taxonomic_names_on_leaves = True
#Read in the x vector
fid = open(input_file, 'r')
x = map(lambda y: float(y), fid.readlines())
fid.close()
#Normalize the x vector
#x = map(lambda y: y/sum(x),x)
#Read in the taxonomy
taxonomy = list()
fid = open(os.path.join(common_kmer_data_path, "Taxonomy.txt"), 'r')
for line in fid:
taxonomy.append(
'_'.join(line.split()[0].split("_")[1:])
) #Just take the first line of the taxonomy (erasing the taxID)
fid.close()
#Read in the basis for the ckm matrices
x_file_names = list()
fid = open(os.path.join(common_kmer_data_path, "FileNames.txt"), 'r')
for line in fid:
x_file_names.append(os.path.basename(line.strip()))
fid.close()
#Read in the common kmer matrix
f = h5py.File(
os.path.join(common_kmer_data_path, 'CommonKmerMatrix-30mers.h5'), 'r')
ckm30 = np.array(f['common_kmers'], dtype=np.float64)
f.close()
f = h5py.File(
os.path.join(common_kmer_data_path, 'CommonKmerMatrix-50mers.h5'), 'r')
ckm50 = np.array(f['common_kmers'], dtype=np.float64)
f.close()
ckm30_norm = np.multiply(ckm30, 1 / np.diag(ckm30))
ckm50_norm = np.multiply(ckm50, 1 / np.diag(ckm50))
num_rows = ckm30_norm.shape[0]
num_cols = ckm30_norm.shape[1]
names = x_file_names
matrix = list()
for i in range(num_rows):
matrix.append([
.5 * (1 - .5 * ckm30_norm[i, j] - .5 * ckm30_norm[j, i]) + .5 *
(1 - .5 * ckm50_norm[i, j] - .5 * ckm50_norm[j, i])
for j in range(i + 1)
])
#Construct the tree. Note I could use RapidNJ here, but a few tests have shown that the trees that RapidNJ creates are rubbish.
dm = _DistanceMatrix(names, matrix)
constructor = DistanceTreeConstructor()
tree = constructor.nj(dm)
t = Tree(tree.format('newick'), format=1)
#tree.format('newick')
#Phylo.draw_ascii(tree)
#Now I will put internal nodes in a certain phylogenetic distance between the root and a given node.
#Function to insert a node at a given distance
def insert_node(t, name_to_insert, insert_above, dist_along):
insert_at_node = t.search_nodes(name=insert_above)[0]
parent = (t & insert_above).up
orig_branch_length = t.get_distance(insert_at_node, parent)
if orig_branch_length < dist_along:
raise ValueError(
"error: dist_along larger than orig_branch_length")
removed_node = insert_at_node.detach()
removed_node.dist = orig_branch_length - dist_along
added_node = parent.add_child(name=name_to_insert, dist=dist_along)
added_node.add_child(removed_node)
#Function to insert a node some % along a branch
def insert_hyp_node(t, leaf_name, percent):
total_dist = t.get_distance(t.name, leaf_name)
percent_dist = percent * total_dist
child_node = (t & leaf_name)
ancestor_node = (t & child_node.name).up
while t.get_distance(t.name, ancestor_node) > percent_dist:
child_node = ancestor_node
ancestor_node = (t & child_node.name).up
insert_node(t, leaf_name + "_" + str(percent), child_node.name,
percent_dist - t.get_distance(t.name, ancestor_node))
#Insert hypothetical nodes
hyp_node_names = dict()
cutoffs = [.9, .8, .7, .6, .5, .4, .3, .2, .1]
cutoffs = map(lambda y: y**1.5, cutoffs)
for i in range(len(x_file_names)):
xi = x[i:len(x):len(x_file_names)]
for j in range(1, len(cutoffs) + 1):
if xi[j] > 0:
insert_hyp_node(t, x_file_names[i], cutoffs[j - 1])
hyp_node_names[x_file_names[i] + "_" + str(cutoffs[j - 1])] = [
x_file_names[i], cutoffs[j - 1], j - 1
] #in case there are "_" in the file names
#insert_hyp_node(t, x_file_names[i],.5/t.get_distance(t.name,t&x_file_names[i])*cutoffs[j])
#Now put the bubbles on the nodes
def layout(node):
#print(node)
if node.is_leaf():
if node.name in x_file_names:
#make reconstructed bubble
size = x[x_file_names.index(node.name)]
F = CircleFace(radius=500 * math.sqrt(size),
color="RoyalBlue",
style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F, node, 0, position="branch-right")
if taxonomic_names_on_leaves:
nameFace = AttrFace(
"name",
fsize=25,
fgcolor='black',
text_suffix="_" +
taxonomy[x_file_names.index(node.name)])
faces.add_face_to_node(nameFace,
node,
0,
position="branch-right")
else:
nameFace = AttrFace("name", fsize=25, fgcolor='black')
faces.add_face_to_node(nameFace,
node,
0,
position="branch-right")
elif node.name in hyp_node_names: #Otherwise it's a hypothetical node, just use recon x
node_base_name = hyp_node_names[node.name][0]
percent = hyp_node_names[node.name][1]
if node_base_name in x_file_names:
idx = hyp_node_names[node.name][2]
size = x[x_file_names.index(node_base_name) +
(idx + 1) * len(x_file_names)]
F = CircleFace(radius=500 * math.sqrt(size),
color="RoyalBlue",
style="sphere")
F.border.width = None
F.opacity = 0.6
faces.add_face_to_node(F, node, 0, position="branch-right")
#print node
#print size
else:
size = 0
else:
size = 0
#print(size)
ts = TreeStyle()
ts.layout_fn = layout
if plot_rectangular:
ts.mode = "r"
else:
ts.mode = "c"
ts.show_leaf_name = False
ts.min_leaf_separation = 50
#Export the tree to a png image
t.render(out_file, w=width, units="mm", tree_style=ts)
#Export the xml file
project = Phyloxml()
phylo = phyloxml.PhyloxmlTree(newick=t.write(format=0, features=[]))
phylo.phyloxml_phylogeny.set_name(title)
project.add_phylogeny(phylo)
project.export(open(out_file_xml, 'w'))