Logo Search packages:      
Sourcecode: python-biopython version File versions  Download package


# Copyright (C) 2009 by Eric Talevich (eric.talevich@gmail.com)
# This code is part of the Biopython distribution and governed by its
# license. Please see the LICENSE file that should have been included
# as part of this package.

"""Base classes for Bio.Phylo objects.

All object representations for phylogenetic trees should derive from these base
classes in order to use the common methods defined on them.
__docformat__ = "epytext en"

import collections
import itertools
import random
import re

import _sugar

# General tree-traversal algorithms

def _level_traverse(root, get_children):
    """Traverse a tree in breadth-first (level) order."""
    Q = collections.deque([root])
    while Q:
        v = Q.popleft()
        yield v

def _preorder_traverse(root, get_children):
    """Traverse a tree in depth-first pre-order (parent before children)."""
    def dfs(elem):
        yield elem
        for v in get_children(elem):
            for u in dfs(v):
                yield u
    for elem in dfs(root):
        yield elem

def _postorder_traverse(root, get_children):
    """Traverse a tree in depth-first post-order (children before parent)."""
    def dfs(elem):
        for v in get_children(elem):
            for u in dfs(v):
                yield u
        yield elem
    for elem in dfs(root):
        yield elem

def _sorted_attrs(elem):
    """Get a flat list of elem's attributes, sorted for consistency."""
    singles = []
    lists = []
    # Sort attributes for consistent results
    for child in sorted(elem.__dict__.itervalues()):
        if child is None:
        if isinstance(child, list):
    return (x for x in singles + lists
            if isinstance(x, TreeElement))

# Factory functions to generalize searching for clades/nodes

def _identity_matcher(target):
    """Match a node to the target object by identity."""
    def match(node):
        return (node is target)
    return match

def _class_matcher(target_cls):
    """Match a node if it's an instance of the given class."""
    def match(node):
        return isinstance(node, target_cls)
    return match

def _attribute_matcher(kwargs):
    """Match a node by specified attribute values.

    'terminal' is a special case: True restricts the search to external (leaf)
    nodes, False restricts to internal nodes, and None allows all tree elements
    to be searched, including phyloXML annotations.

    Otherwise, for a tree element to match the specification (i.e. for the
    function produced by _attribute_matcher to return True when given a tree
    element), it must have each of the attributes specified by the keys and
    match each of the corresponding values -- think 'and', not 'or', for
    multiple keys.
    def match(node):
        if 'terminal' in kwargs:
            # Special case: restrict to internal/external/any nodes
            kwa_copy = kwargs.copy()
            pattern = kwa_copy.pop('terminal')
            if (pattern is not None and
                (not hasattr(node, 'is_terminal') or
                    node.is_terminal() != pattern)):
                return False
            kwa_copy = kwargs
        for key, pattern in kwa_copy.iteritems():
            # Nodes must match all other specified attributes
            if not hasattr(node, key):
                return False
            target = getattr(node, key)
            if isinstance(pattern, basestring):
                return (isinstance(target, basestring) and
                        re.match(pattern+'$', target))
            if isinstance(pattern, bool):
                return (pattern == bool(target))
            if isinstance(pattern, int):
                return (pattern == target)
            if pattern is None:
                return (target is None)
            raise TypeError('invalid query type: %s' % type(pattern))
        return True
    return match

def _function_matcher(matcher_func):
    """Safer attribute lookup -- returns False instead of raising an error."""
    def match(node):
            return matcher_func(node)
        except (LookupError, AttributeError, ValueError):
            return False
    return match

def _object_matcher(obj):
    """Retrieve a matcher function by passing an arbitrary object.

    i.e. passing a TreeElement such as a Node or Tree instance returns an
    identity matcher, passing a type such as the PhyloXML.Taxonomy class returns
    a class matcher, and passing a dictionary returns an attribute matcher.

    The resulting 'match' function returns True when given an object matching
    the specification (identity, type or attribute values), otherwise False.
    This is useful for writing functions that search the tree, and probably
    shouldn't be used directly by the end user.
    if isinstance(obj, TreeElement):
        return _identity_matcher(obj)
    if isinstance(obj, type):
        return _class_matcher(obj)
    if isinstance(obj, dict):
        return _attribute_matcher(obj)
    if callable(obj):
        return _function_matcher(obj)
    raise ValueError("%s (type %s) is not a valid type for comparison."
                     % (obj, type(obj)))

def _combine_matchers(target, kwargs, require_spec):
    if not target:
        if not kwargs:
            if require_spec:
                raise ValueError("you must specify a target object or keyword "
            return lambda x: True
        return _attribute_matcher(kwargs)
    match_obj = _object_matcher(target)
    if not kwargs:
        return match_obj
    match_kwargs = _attribute_matcher(kwargs)
    return (lambda x: match_obj(x) and match_kwargs(x))

# Class definitions

00170 class TreeElement(object):
    """Base class for all Bio.Phylo classes."""

00173     def __repr__(self):
        """Show this object's constructor with its primitive arguments."""
        def pair_as_kwarg_string(key, val):
            if isinstance(val, basestring):
                return "%s='%s'" % (key, _sugar.trim_str(unicode(val)))
            return "%s=%s" % (key, val)
        s = '%s(%s)' % (self.__class__.__name__,
                        ', '.join(pair_as_kwarg_string(key, val)
                                  for key, val in self.__dict__.iteritems()
                                  if val is not None and
                                  type(val) in (str, int, float, bool, unicode)
        return s.encode('utf-8')

    __str__ = __repr__

00190 class TreeMixin(object):
    """Methods for Tree- and Clade-based classes.

    This lets Tree and Clade support the same traversal and searching
    operations without requiring Clade to inherit from Tree, so Clade isn't
    required to have all of Tree's attributes -- just 'root' (a Clade
    instance) and 'is_terminal()'.
    # Traversal methods

00200     def _filter_search(self, filter_func, order, follow_attrs):
        """Perform a BFS or DFS traversal through all elements in this tree.

        @return: generator of all elements for which 'filter_func' is True.
        order_opts = {'preorder': _preorder_traverse,
                      'postorder': _postorder_traverse,
                      'level': _level_traverse}
            order_func = order_opts[order]
        except KeyError:
            raise ValueError("Invalid order '%s'; must be one of: %s"
                             % (order, tuple(order_opts.keys())))
        if follow_attrs:
            get_children = _sorted_attrs
            root = self
            get_children = lambda elem: elem.clades
            root = self.root
        return itertools.ifilter(filter_func, order_func(root, get_children))

00221     def find_any(self, *args, **kwargs):
        """Return the first element found by find_elements(), or None.

        This is also useful for checking whether any matching element exists in
        the tree.
        hits = self.find_elements(*args, **kwargs)
            return hits.next()
        except StopIteration:
            return None

00233     def find_elements(self, target=None, terminal=None, order='preorder',
        """Find all tree elements matching the given attributes.

        The arbitrary keyword arguments indicate the attribute name of the
        sub-element and the value to match: string, integer or boolean. Strings
        are evaluated as regular expression matches; integers are compared
        directly for equality, and booleans evaluate the attribute's truth value
        (True or False) before comparing. To handle nonzero floats, search with
        a boolean argument, then filter the result manually.

        If no keyword arguments are given, then just the class type is used for

        The result is an iterable through all matching objects, by depth-first
        search. (Not necessarily the same order as the elements appear in the
        source file!)


            >>> from Bio.Phylo.IO import PhyloXMIO
            >>> phx = PhyloXMLIO.read('phyloxml_examples.xml')
            >>> matches = phx.phylogenies[5].find_elements(code='OCTVU')
            >>> matches.next()
            Taxonomy(code='OCTVU', scientific_name='Octopus vulgaris')

        @param target: 
            Specifies the characteristics to search for. (The default,
            TreeElement, matches any standard Bio.Phylo type.)
        @type target: TreeElement instance, type, dict, or callable

        @param terminal:
            A boolean value to select for or against terminal nodes (a.k.a. leaf
            nodes). True searches for only terminal nodes, False excludes
            terminal nodes, and the default, None, searches both terminal and
            non-terminal nodes, as well as any tree elements lacking the
            'is_terminal' method.
        @type terminal: bool

        @param order:
            Tree traversal order: 'preorder' (default) is depth-first search,
            'postorder' is DFS with child nodes preceding parents, and 'level'
            is breadth-first search.
        @type order: string ('preorder'|'postorder'|'level')
        if terminal is not None:
            kwargs['terminal'] = terminal
        is_matching_elem = _combine_matchers(target, kwargs, False)
        return self._filter_search(is_matching_elem, order, True)

00283     def find_clades(self, target=None, terminal=None, order='preorder',
        """Find each clade containing a matching element.

        That is, find each element as with find_elements(), but return the
        corresponding clade object.
        def match_attrs(elem):
            orig_clades = elem.__dict__.pop('clades')
            found = elem.find_any(target, **kwargs)
            elem.clades = orig_clades
            return (found is not None)
        if terminal is None:
            is_matching_elem = match_attrs
            def is_matching_elem(elem):
                return ((elem.is_terminal() == terminal) and
        return self._filter_search(is_matching_elem, order, False)

00303     def get_path(self, target=None, **kwargs):
        """List the clades directly between the root and the given target.

        Returns a list of all clade objects along this path, ending with
        the given target, but excluding the root clade.
        # Only one path will work -- ignore weights and visits
        path = []
        match = _combine_matchers(target, kwargs, True)
        def check_in_path(v):
            if match(v):
                return True
            elif v.is_terminal():
                return False
            for child in v:
                if check_in_path(child):
                    return True
            return False
        if not check_in_path(self.root):
            return None
        return path[-2::-1]

00327     def get_nonterminals(self, order='preorder'):
        """Get a list of all of this tree's nonterminal (internal) nodes."""
        return list(self.find_clades(terminal=False, order=order))

00331     def get_terminals(self, order='preorder'):
        """Get a list of all of this tree's terminal (leaf) nodes."""
        return list(self.find_clades(terminal=True, order=order))

00335     def trace(self, start, finish):
        """List of all clade object between two targets in this tree.

        Excluding start, including finish.
        mrca = self.common_ancestor(start, finish)
        fromstart = mrca.get_path(start)[-2::-1]
        to = mrca.get_path(finish)
        return fromstart + [mrca] + to

    # Information methods

00347     def common_ancestor(self, *targets):
        """Most recent common ancestor (clade) of all the given targets.

        Edge cases: 

            - If no target is given, returns self.root
            - If 1 target is given, returns the target
            - If any target is not found in this tree, raises a ValueError
        paths = [self.get_path(t) for t in targets]
        # Validation -- otherwise izip throws a spooky error below
        for p, t in zip(paths, targets):
            if p is None:
                raise ValueError("target %s is not in this tree" % repr(t))
        mrca = self.root
        for level in itertools.izip(*paths):
            ref = level[0]
            for other in level[1:]:
                if ref is not other:
                mrca = ref
            if ref is not mrca:
        return mrca

00373     def count_terminals(self):
        """Counts the number of terminal (leaf) nodes within this tree."""
        return _sugar.iterlen(self.find_clades(terminal=True))

00377     def depths(self, unit_branch_lengths=False):
        """Create a mapping of tree clades to depths (by branch length).

        @return: dict of {clade: depth}
        if unit_branch_lengths:
            depth_of = lambda c: 1
            depth_of = lambda c: c.branch_length or 0
        depths = {}
        def update_depths(node, curr_depth):
            depths[node] = curr_depth
            for child in node.clades:
                new_depth = curr_depth + depth_of(child)
                update_depths(child, new_depth)
        update_depths(self.root, 0)
        return depths

00395     def distance(self, target1, target2=None):
        """Calculate the sum of the branch lengths between two targets.

        If only one target is specified, the other is the root of this tree.
        if target2 is None:
            return sum(n.branch_length for n in self.get_path(target1)
                       if n.branch_length is not None)
        mrca = self.common_ancestor(target1, target2)
        return mrca.distance(target1) + mrca.distance(target2)

00406     def is_bifurcating(self):
        """Return True if tree downstream of node is strictly bifurcating."""
        # Root can be trifurcating, because it has no ancestor
        if isinstance(self, Tree) and len(self.root) == 3:
            return (self.root.clades[0].is_bifurcating() and
                    self.root.clades[1].is_bifurcating() and
        if len(self.root) == 2:
            return (self.root.clades[0].is_bifurcating() and
        if len(self.root) == 0:
            return True
        return False

00420     def is_monophyletic(self, terminals):
        """MRCA of terminals if they comprise a complete subclade, or False.

        @return: common ancestor if terminals are monophyletic, otherwise False.
        target_set = set(terminals)
        current = self.root
        while True:
            if set(current.get_terminals()) == target_set:
                return current
            # Try a narrower subclade
            for subclade in current.clades:
                if set(subclade.get_terminals()).issuperset(target_set):
                    current = subclade
                return False

00438     def is_parent_of(self, target=None, **kwargs):
        """True if target is a descendent of this tree.

        Not required to be a direct descendent.
        return (self.get_path(target, **kwargs) is not None)

00445     def is_preterminal(self):
        """True if all direct descendents are terminal."""
        if self.root.is_terminal():
            return False
        for clade in self.root.clades:
            if not clade.is_terminal():
                return False
        return True

00454     def total_branch_length(self):
        """Calculate the sum of all the branch lengths in this tree."""
        return sum(node.branch_length
                   for node in self.find_clades(branch_length=True))

    # Tree manipulation methods

00461     def collapse(self, target=None, **kwargs):
        """Deletes target from the tree, relinking its children to its parent.

        @return: the parent clade.
        path = self.get_path(target, **kwargs)
        if not path:
            raise ValueError("couldn't collapse %s in this tree"
                             % (target or kwargs))
        if len(path) == 1:
            parent = self.root
            parent = path[-2]
        popped = parent.clades.pop(parent.clades.index(path[-1]))
        extra_length = popped.branch_length or 0
        for child in popped:
            child.branch_length += extra_length
        return parent

00481     def collapse_all(self):
        """Collapse all the descendents of this tree, leaving only terminals.

        To collapse only certain elements, use the collapse method directly in a
        loop with find_clades:

        >>> for clade in tree.find_clades(branch_length=True, order='level'):
        >>>     if (clade.branch_length < .5 and
        >>>         not clade.is_terminal() and
        >>>         clade is not self.root):
        >>>         tree.collapse(clade)

        Note that level-order traversal helps avoid strange side-effects when
        modifying the tree while iterating over its clades.
        internals = self.find_clades(terminal=False, order='level')
        # Skip the root node -- it can't be collapsed
        for clade in internals:

00502     def ladderize(self, reverse=False):
        """Sort clades in-place according to the number of terminal nodes.

        Deepest clades are last by default. Use reverse=True to sort clades
        self.root.clades.sort(key=lambda c: c.count_terminals(),
        for subclade in self.root.clades:

00513     def prune(self, target=None, **kwargs):
        """Prunes a terminal clade from the tree.

        If taxon is from a bifurcation, the connecting node will be collapsed
        and its branch length added to remaining terminal node. This might be no
        longer a meaningful value.

        @return: parent clade of the pruned target
        if 'terminal' in kwargs and kwargs['terminal']:
            raise ValueError("target must be terminal")
        path = self.get_path(target, terminal=True, **kwargs)
        if not path:
            raise ValueError("can't find a matching target below this root")
        if len(path) == 1:
            parent = self.root
            parent = path[-2]
        if len(parent) == 1:
            # We deleted a branch from a bifurcation
            if parent == self.root:
                # If we're at the root, move the root upwards
                # NB: This loses the length of the original branch
                newroot = parent.clades[0]
                newroot.branch_length = None
                parent = self.root = newroot
                # If we're not at the root, collapse this parent
                child = parent.clades[0]
                if child.branch_length is not None:
                    child.branch_length += (parent.branch_length or 0.0)
                if len(path) < 3:
                    grandparent = self.root
                    grandparent = path[-3]
                # Replace parent with child at the same place in grandparent
                index = grandparent.clades.index(parent)
                grandparent.clades.insert(index, child)
                parent = grandparent
        return parent

00556     def split(self, n=2, branch_length=1.0):
        """Speciation: generate n (default 2) new descendants.

        New clades have the given branch_length and the same name as this
        clade's root plus an integer suffix (counting from 0).
        subtree_cls = type(self.root)
        base_name = self.root.name or ''
        for i in range(n):
            clade = subtree_cls(name=base_name+str(i),

00570 class Tree(TreeElement, TreeMixin):
    """A phylogenetic tree, containing global info for the phylogeny.

    The structure and node-specific data is accessible through the 'root'
    subtree attached to the Tree instance.

    @param root:
        The starting node of the tree. If the tree is rooted, this will usually
        be the root node.
    @type root: Clade

    @param rooted:
        Whether or not the tree is rooted. By default, a tree is assumed to be
    @type rooted: bool

    @param id: The identifier of the tree, if there is one.
    @type id: str

    @param name: The name of the tree, in essence a label.
    @type name: str
    def __init__(self, root=None, rooted=True, id=None, name=None):
        self.root = root or Clade()
        self.rooted = rooted
        self.id = id
        self.name = name

00599     def from_subtree(cls, subtree, **kwargs):
        """Create a new Tree object given a subtree.

        Keyword arguments are the usual Tree constructor parameters.
        return cls(subtree, **kwargs)

00607     def randomized(cls, taxa, branch_length=1.0, branch_stdev=None):
        """Create a randomized bifurcating tree given a list of taxa.

        @param taxa: Either an integer specifying the number of taxa to create
            (automatically named taxon#), or an iterable of taxon names, as

        @return: a tree of the same type as this class.
        if isinstance(taxa, int):
            taxa = ['taxon%s' % (i+1) for i in xrange(taxa)]
        elif hasattr(taxa, '__iter__'):
            taxa = list(taxa)
            raise TypeError("taxa argument must be integer (# taxa) or "
                            "iterable of taxon names.")
        rtree = cls()
        terminals = [rtree.root]
        while len(terminals) < len(taxa):
            newsplit = random.choice(terminals)
            newterms = newsplit.split(branch_length=branch_length)
            if branch_stdev:
                # Add some noise to the branch lengths
                for nt in newterms:
                    nt.branch_length = max(0,
                            random.gauss(branch_length, branch_stdev))
        # Distribute taxon labels randomly
        for node, name in zip(terminals, taxa):
            node.name = name
        return rtree

00642     def clade(self):
        """The first subtree in this tree (not itself)."""
        return self.root

    # Method assumed by TreeMixin

00648     def is_terminal(self):
        """True if the root of this tree is terminal."""
        return (not self.root.clades)

    # Convention from SeqRecord and Alignment classes  

00654     def __format__(self, format_spec):
        """Serialize the tree as a string in the specified file format.

        This method supports the format() built-in function added in Python
        2.6/3.0. The format_spec should be a lower case string supported by
        Bio.Phylo.write as an output file format. 
        if format_spec:
            from StringIO import StringIO
            import _io
            handle = StringIO()
            _io.write([self], handle, format_spec)
            return handle.getvalue()
            # Follow python convention and default to using __str__
            return str(self)

00671     def format(self, format):
        """Serialize the tree as a string in the specified file format.

        This duplicates the __format__ magic method for pre-2.6 Pythons.
        return self.__format__(format)

    # Pretty-printer for the entire tree hierarchy

00680     def __str__(self):
        """String representation of the entire tree.

        Serializes each sub-clade recursively using repr() to create a summary
        of the object structure.
        TAB = '    '
        textlines = []
        def print_tree(obj, indent):
            """Recursively serialize sub-elements.

            This closes over textlines and modifies it in-place.
            textlines.append(TAB*indent + repr(obj))
            indent += 1
            for attr in obj.__dict__:
                child = getattr(obj, attr)
                if isinstance(child, TreeElement):
                    print_tree(child, indent)
                elif isinstance(child, list):
                    for elem in child:
                        if isinstance(elem, TreeElement):
                            print_tree(elem, indent)
        print_tree(self, 0)
        return '\n'.join(textlines)

00707 class Clade(TreeElement, TreeMixin):
    """A recursively defined subtree.

    @param branch_length:
        The length of the branch leading to the root node of this subtree.
    @type branch_length: str

    @param name: The clade's name (a label).
    @type name: str

    @param clades: Sub-trees rooted directly under this tree's root.
    @type clades: list
    def __init__(self, branch_length=None, name=None, clades=None):
        self.clades = clades or []
        self.name = name
        self.branch_length = branch_length

00726     def root(self):
        """Allow TreeMixin methods to traverse subtrees properly."""
        return self

00730     def is_terminal(self):
        """True if this is a terminal (leaf) node."""
        return (not self.clades)

    # Sequence-type behavior methods

00736     def __getitem__(self, index):
        """Get subtrees by index (integer or slice)."""
        if isinstance(index, int) or isinstance(index, slice):
            return self.clades[index]
        ref = self
        for idx in index:
            ref = ref[idx]
        return ref

00745     def __iter__(self):
        """Iterate through this tree's direct subtrees (clades)."""
        return iter(self.clades)

00749     def __len__(self):
        """Number of subtrees directy under the root."""
        return len(self.clades)

00753     def __nonzero__(self):
        """Boolean value of an instance of this class.

        NB: If this method is not defined, but __len__  is, then the object is
        considered true if the result of __len__() is nonzero. We want Clade
        instances to always be considered true.
        return True

    def __str__(self):
        if self.name:
            return _sugar.trim_str(self.name, maxlen=40)
        return self.__class__.__name__

Generated by  Doxygen 1.6.0   Back to index