# Only write a new alignment if there is a new alignment
if empty_seq_count >= 1:
if len(trimmed_alignment) >= 1:
SeqIO.write(trimmed_alignment, handle=alignment_file, \
format=alignment_format)
tree = EvolTree(tree_file)
out_tree_name = os.path.basename(tree_file)
out_tree_name = os.path.splitext(out_tree_name)[0]
out_tree_name = out_tree_name + '_' + gene_name + '.tre'
# If there is a new alignment, prune the tree down to the taxa that remain in
# the new alignment
if empty_seq_count >= 1:
if len(taxa_in_alignment) >= 1:
tree.prune(taxa_in_alignment, preserve_branch_length=True)
test_taxa = []
with open(test_taxa_file, 'r') as test_taxa_list:
for taxon in test_taxa_list:
taxon = taxon.rstrip()
test_taxa.append(taxon)
nodes_to_mark = set() # set since we want it to be all unique ids
# Mark the test taxa
for taxon in test_taxa:
taxon_node = tree & taxon # ete3 notation for finding a node within a tree
taxon_id = taxon_node.node_id # mark_tree only takes node_ids, not labels
nodes_to_mark.add(taxon_id)
