#!/usr/bin/env python
'''Reads OTT files and produces a set of files with preorder
numbering that allow of minimal memory footprint in python scripts
Note: this script is a memory hog
'''
if __name__ == '__main__':
from peyotl.ott import OTT
ott = OTT()
ott.create_pickle_files()
out = codecs.open(outfn, mode='w', encoding='utf-8')
except:
sys.exit('validate_ot_nexson: Could not open output filepath "{fn}"\n'.format(fn=outfn))
else:
out = codecs.getwriter('utf-8')(sys.stdout)
try:
nexson = read_as_json(inp_filepath)
except ValueError as vx:
_LOG.error('Not valid JSON.')
if args.verbose:
raise vx
else:
sys.exit(1)
except Exception as nx:
_LOG.error(nx.value)
sys.exit(1)
convert_nexson_format(nexson, BY_ID_HONEY_BADGERFISH)
trees = extract_tree_nexson(nexson, tree_id=args.tree_id)
if len(trees) == 0:
trees = extract_tree_nexson(nexson, tree_id=None)
if trees:
v = '", "'.join([i[0] for i in trees])
sys.exit('Tree ID {i} not found. Valid IDs for this file are "{l}"\n'.format(i=args.tree_id, l=v))
else:
sys.exit('This NexSON has not trees.\n')
ott = OTT()
for tree_id, tree, otus in trees:
tree_proxy = NexsonTreeProxy(tree=tree, tree_id=tree_id, otus=otus)
evaluate_tree_rooting(nexson, ott, tree_proxy)
required=False,
help='Optional comma-separated list of flags to prune. '
'If omitted, the treemachine flags are used.')
parser.add_argument('--root',
default=None,
type=int,
required=False,
help='Optional taxonomy root argument.')
args = parser.parse_args(sys.argv[1:])
ott_dir, output, log_filename, root = args.ott_dir, args.output, args.log, args.root
flags_str = args.flags
try:
assert os.path.isdir(args.ott_dir)
except:
sys.exit('Expecting ott-dir argument to be a directory. Got "{}"'.format(args.ott_dir))
ott = OTT(ott_dir=args.ott_dir)
if flags_str is None:
flags = ott.TREEMACHINE_SUPPRESS_FLAGS
else:
flags = flags_str.split(',')
create_log = log_filename is not None
with codecs.open(args.output, 'w', encoding='utf-8') as outp:
log = ott.write_newick(outp,
label_style=OTULabelStyleEnum.CURRENT_LABEL_OTT_ID,
root_ott_id=root,
prune_flags=flags,
create_log_dict=create_log)
outp.write('\n')
if create_log:
write_as_json(log, log_filename)
inp_files = [i.strip() for i in tf if i.strip()]
else:
error(
'nexson file must be specified as a positional argument or via the --nexson-file-tags or --input-files-list argument.')
sys.exit(1)
if not inp_files:
error('No input files specified.')
in_dir = args.input_dir
if in_dir:
in_dir = os.path.expanduser(in_dir)
inp_files = [os.path.join(in_dir, i) for i in inp_files]
if flags_str is None:
flags = OTT.TREEMACHINE_SUPPRESS_FLAGS
else:
flags = flags_str.split(',')
ott = OTT(ott_dir=args.ott_dir)
to_prune_fsi_set = ott.convert_flag_string_set_to_union(flags)
for inp in inp_files:
_LOG.debug('{}'.format(inp))
log_obj = {}
inp_fn = os.path.split(inp)[-1]
study_tree = '.'.join(inp_fn.split('.')[:-1]) # strip extension
study_id, tree_id = propinquity_fn_to_study_tree(inp_fn)
nexson_blob = read_as_json(inp)
ntw = NexsonTreeWrapper(nexson_blob, tree_id, log_obj=log_obj)
assert ntw.root_node_id
taxonomy_treefile = os.path.join(args.out_dir, study_tree + '-taxonomy.tre')
try:
ntw.prune_tree_for_supertree(ott=ott,
to_prune_fsi_set=to_prune_fsi_set,
root_ott_id=root,
'If omitted, the treemachine flags are used.')
parser.add_argument('--root',
default=None,
type=int,
required=False,
help='Optional taxonomy root argument.')
args = parser.parse_args(sys.argv[1:])
ott_dir, output, log_filename, root = args.ott_dir, args.output, args.log, args.root
flags_str = args.flags
try:
assert os.path.isdir(args.ott_dir)
except:
sys.exit(
'Expecting ott-dir argument to be a directory. Got "{}"'.format(
args.ott_dir))
ott = OTT(ott_dir=args.ott_dir)
if flags_str is None:
flags = ott.TREEMACHINE_SUPPRESS_FLAGS
else:
flags = flags_str.split(',')
create_log = log_filename is not None
with codecs.open(args.output, 'w', encoding='utf-8') as outp:
log = ott.write_newick(
outp,
label_style=OTULabelStyleEnum.CURRENT_LABEL_OTT_ID,
root_ott_id=root,
prune_flags=flags,
create_log_dict=create_log)
outp.write('\n')
if create_log:
write_as_json(log, log_filename)
def ott_shell_command(args):
ott = OTT()
_LOG.info('launching bash in your OTT dir...')
if subprocess.Popen('bash', cwd=ott.ott_dir).wait() != 0:
raise RuntimeError('bash in ott dir failed.')
def ott_clear_command(args):
ott = OTT()
ott.remove_caches()
else:
error(
'nexson file must be specified as a positional argument or via the --nexson-file-tags or --input-files-list argument.'
)
sys.exit(1)
if not inp_files:
error('No input files specified.')
in_dir = args.input_dir
if in_dir:
in_dir = os.path.expanduser(in_dir)
inp_files = [os.path.join(in_dir, i) for i in inp_files]
if flags_str is None:
flags = OTT.TREEMACHINE_SUPPRESS_FLAGS
else:
flags = flags_str.split(',')
ott = OTT(ott_dir=args.ott_dir)
to_prune_fsi_set = ott.convert_flag_string_set_to_union(flags)
for inp in inp_files:
_LOG.debug('{}'.format(inp))
log_obj = {}
inp_fn = os.path.split(inp)[-1]
study_tree = '.'.join(inp_fn.split('.')[:-1]) # strip extension
study_id, tree_id = propinquity_fn_to_study_tree(inp_fn)
nexson_blob = read_as_json(inp)
ntw = NexsonTreeWrapper(nexson_blob, tree_id, log_obj=log_obj)
assert ntw.root_node_id
taxonomy_treefile = os.path.join(args.out_dir,
study_tree + '-taxonomy.tre')
try:
ntw.prune_tree_for_supertree(
ott=ott,