def test_majority_consensus(self):
ref_trees = Phylo.parse('./TreeConstruction/majority_ref.tre', 'newick')
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees, 1)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
def test_majority_consensus(self):
ref_trees = Phylo.parse('./TreeConstruction/majority_ref.tre', 'newick')
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees, 1)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
def test_majority_consensus(self):
# three trees
# ref_tree = open('./TreeConstruction/majority_ref.tre')
ref_tree = list(Phylo.parse("./TreeConstruction/majority_ref.tre", "newick"))
consensus_tree = Consensus.majority_consensus(self.trees)
# tree_file = StringIO()
# Phylo.write(consensus_tree, tree_file, 'newick')
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree[0]))
consensus_tree = Consensus.majority_consensus(self.trees, 1)
# tree_file = StringIO()
# Phylo.write(consensus_tree, tree_file, 'newick')
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree[1]))
def test_majority_consensus(self):
ref_trees = Phylo.parse("./TreeConstruction/majority_ref.tre", "newick")
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
# ref_tree = next(ref_trees)
# consensus_tree_mcmc = Consensus.majority_consensus(self.mcmc_trees, mcmc=True)
# self.assertTrue(Consensus._equal_topology(consensus_tree_mcmc, ref_tree))
ref_tree = next(ref_trees)
consensus_tree = Consensus.majority_consensus(self.trees, 1)
self.assertTrue(Consensus._equal_topology(consensus_tree, ref_tree))
consensus_tree_mcmc = Consensus.majority_consensus(self.mcmc_trees, 1, mcmc=True)
self.assertTrue(Consensus._equal_topology(consensus_tree_mcmc, ref_tree))
def _calculate_gsi(self):
"""
Method for calculating Gene Support Indices
:return:
"""
LOGGER.info("Calculating Gene Support Indices (GSIs)"
" from the gene trees..")
genome_num = 0
bcg_dir = os.path.join(self._dirpath, self.config.bcg_dir)
for file in os.listdir(bcg_dir):
if file.endswith('.bcg'):
genome_num += 1
nwk_file = os.path.join(self._align_output_dir, "all_gene.trees")
trees = Phylo.parse(nwk_file, 'newick')
tree = Consensus.majority_consensus(trees,
cutoff=(100-self.config.gsi_threshold) * genome_num/100)
Phylo.draw_ascii(tree)
ubcg_gsi_file = os.path.join(self._align_output_dir,
f'UBCG_gsi({self._bcg_num}'
f'){self.config.postfixes.align_tree_const}')
with open(ubcg_gsi_file, 'w') as f:
Phylo.write(tree, f, 'newick')
LOGGER.info("The final tree marked with GSI was written"
" to %s", ubcg_gsi_file)
def bootstrap(self, afbased=True, basename='majorityTree', treebuilder='nj', bootstraps=1000, outgroup=None, useAllLoci=False):
"""treebuilder could be nj/upgma, outgroup: a population name or 'midpoint'"""
## allLoci: all loci that are variable in at least one population
allpolySites, pwm = {True: ['allpolySites', 'pwm'],
False: ['allpolySitesVCF', 'pwmVCF']}[afbased]
allLoci = set()
for pop in self.populations:
allLoci = allLoci.union(getattr(pop, allpolySites))
allLoci = list(allLoci) ## sort it? Reduce to independent sites (www.pnas.org/content/93/23/13429, run LD?)
sites = len(allLoci)
trees = []
print ("Bootstrapping, rounds:", end=' ')
for bootstrap in range(bootstraps): ## see also parallelized version
print(bootstrap, end=' ')
if useAllLoci:
selectedLoci = allLoci
else:
selectedLoci0 = np.random.choice(range(len(allLoci)), sites, replace=True)
selectedLoci = [allLoci[l] for l in selectedLoci0]
df = pd.DataFrame([pop.bootstrap(selectedLoci, afbased) for pop in self.populations], index=self.popnames)
#import pdb; pdb.set_trace()
self.dmNei = neiDF(df, [5]*(sites-1))
## annoying conversion, BioPython couldnt be just more compatible with scipy/pdist?
dmTriangular = [list(self.dmNei[i, :(i + 1)]) for i in range(len(self.dmNei))]
m = _DistanceMatrix(self.popnames, dmTriangular)
constructor = DistanceTreeConstructor() # could've passed treebuilder here too
tree = getattr(constructor, treebuilder)(m)
if outgroup == 'midpoint':
tree.root_at_midpoint()
elif not outgroup is None:
tree.root_with_outgroup({'name': outgroup})
trees.append(tree) ## use nj!
## debug info:
print(f'selectedLoci: {selectedLoci[:30]}')
## see https://biopython.org/wiki/Phylo, turned out to be more suitable than dendropy/sumtrees
self.majorityTree = Consensus.majority_consensus(trees) ## also consider strict_consensus and adam_consensus (but they don't have bootstrap support values)
treefile = '%s/%s_%s_%s_%s.nwk' %(resultDir, basename, bootstraps,treebuilder, len(self.populations))
Phylo.write(self.majorityTree, treefile, format='newick')
print(f'wrote {treefile}')
Phylo.draw_ascii(self.majorityTree)
the script parallelBootstrap.py, which dumps a list of trees that were created with BioPython's
DistanceTreeConstructor.
Alternative consensus construction that BioPython provides:
* strict_consensus
* adam_consensus
They don't provide bootstrap support values, though.
#run on HPC
#source activate bio3
"""
datadir = "/research/btc_bioinformatic/operations/HLA/FreqRT/Data"
treebase = "maj_ABC_1_nj_92"
if len(sys.path) > 2:
treebase = sys.argv[-1]
#maj_AB_min95_1_nj_246_0001.pcl
treefiles = glob.glob("%s/%s*.pcl" % (datadir, treebase))
trees = []
for treefile in treefiles:
with open(treefile, "rb") as tf:
trees += pickle.load(tf)
majorityTree = Consensus.majority_consensus(trees)
Phylo.write(majorityTree, '../Data/%s.nwk' % treebase, format='newick')
Phylo.draw_ascii(majorityTree)
