Relative evolutionary divergence (RED)

[ryneches]

Hey folks!

I plotted relative evolutionary divergence in ToyTree. I thought I'd and share gather critiques of my implementation. Eventually, I would like to port over some of the features of PhyloRank into ToyTree, so that RED can be used to place taxon labels onto trees.

What do you think? If I make a PR, should it live in toytree.pcm.diversification? Start a new subpackage?

t = toytree.tree( 'test.nwk' )
toytree.mod.root_on_minimal_ancestor_deviation( t )

RED = {}

for node in t.traverse( strategy='preorder' ) :
    if node.is_root() :
        RED[ node ] = 0
        continue
    if node.is_leaf() :
        RED[ node ] = 1
        continue
    parent = node.get_ancestors()[0]
    P = RED[ parent ]
    a = t.distance.get_node_distance( node, parent )
    b = numpy.mean( [ d for node, d in t.distance.iter_descendant_dists( node ) if node.is_leaf() ] )
    if a + b == 0 :
        raise Exception( 'node {n} : a={a}, b={b}'.format( n=node, a=a, b=b ) )
    RED[ node ] = P + ( a / ( a + b ) ) * ( 1 - P )

t.set_node_data( feature='RED', data=RED, inplace=True )

mystyle = {
    'layout'     : 'd',
    'edge_type'  : 'p',
    'edge_style' : {
        'stroke' : toytree.color.COLORS1[1],
        'stroke-width' : 2.5,
    },
    'tip_labels'  : False,
    'node_labels' : False,
    'node_sizes'  : 10,
    'node_colors' : ( 'RED', 'Reds', 1, 0 )
}

canvas, axes, mark = t.draw( height=600, width=1200, **mystyle )
canvas

[eaton-lab]

Hey Russel, This looks great. It would be awesome for you to implement some of these methods. I think `toytree.pcm` is the right place for it. You could create a new module within there for the phylorank related functions. Here is a slightly streamlined version of the red function that gets a distance matrix up front, and uses .up to reference parents, both of which will increase speed a bit. ```python def get_relative_evolutionary_divergence(tree): """Returns a dict mapping Node idx to relative evolutionary divergence. """ # store root to 0, tips to 1 red = {tree.treenode.idx: 0} red.update({i: 1 for i in range(tree.ntips)}) # get all-by-all node distance matrix mat = tree.distance.get_node_distance_matrix(topology_only=False) # preorder (parent then child) excluding root and tips for node in tree[-2:tree.ntips - 1:-1]: # get parent's red value (P), and dist to parent (a) P = red[node.up.idx] a = mat[node.idx, node.up.idx] # get avg dist from this node to each of its leaves (b) b = np.mean([mat[node.idx, leaf.idx] for leaf in node.iter_leaves()]) # why can the dist from node to parent not be the same as avg to its leaves? if a + b == 0: raise ValueError(f"node {node}: a == b == {a}") # store this nodes red value red[node.idx] = P + (a / (a + b)) * (1 - P) return red ``` Best, Deren

…

On Tue, Jul 30, 2024 at 4:34 AM Russell Neches ***@***.***> wrote: Hey folks! I plotted relative evolutionary divergence <https://www.nature.com/articles/s41564-021-00918-8> in ToyTree. I thought I'd and share gather critiques of my implementation. Eventually, I would like to port over some of the features of PhyloRank <https://github.com/donovan-h-parks/PhyloRank> into ToyTree, so that RED can be used to place taxon labels onto trees. What do you think? If I make a PR, should it live in toytree.pcm.diversification? Start a new subpackage? t = toytree.tree( 'test.nwk' )toytree.mod.root_on_minimal_ancestor_deviation( t ) RED = {} for node in t.traverse( strategy='preorder' ) : if node.is_root() : RED[ node ] = 0 continue if node.is_leaf() : RED[ node ] = 1 continue parent = node.get_ancestors()[0] P = RED[ parent ] a = t.distance.get_node_distance( node, parent ) b = numpy.mean( [ d for node, d in t.distance.iter_descendant_dists( node ) if node.is_leaf() ] ) if a + b == 0 : raise Exception( 'node {n} : a={a}, b={b}'.format( n=node, a=a, b=b ) ) RED[ node ] = P + ( a / ( a + b ) ) * ( 1 - P ) t.set_node_data( feature='RED', data=RED, inplace=True ) mystyle = { 'layout' : 'd', 'edge_type' : 'p', 'edge_style' : { 'stroke' : toytree.color.COLORS1[1], 'stroke-width' : 2.5, }, 'tip_labels' : False, 'node_labels' : False, 'node_sizes' : 10, 'node_colors' : ( 'RED', 'Reds', 1, 0 ) } canvas, axes, mark = t.draw( height=600, width=1200, **mystyle )canvas 77-849_species_RED.svg (view on web) <https://github.com/user-attachments/assets/bd6eb6f0-3134-4fbc-99ec-67e86e63847a> — Reply to this email directly, view it on GitHub <#90>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGEVUG4D4CDIV62TA6325ILZO5FYFAVCNFSM6AAAAABLV3PEGCVHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZWHE4TGOBVGQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Deren Eaton Associate Professor, Department of E3B Affiliate Member, Data Science Institute Committee on Equity and Diversity Columbia University https://eaton-lab.org <http://eaton-lab.org>

[ryneches]

Awesome! I submitted this as PR #91.