picea

Lightweight python library for working with trees and sequence collections

pip install picea

The above figure can be generated with the following code

from picea import Tree
import matplotlib.pyplot as plt

newick = '(((a,b),(c,d)),e)'
tree = Tree.from_newick(newick)

fig, (ax1, ax2) = plt.subplots(ncols = 2, figsize = (10, 4))

#left-to-right layout with direct links
tree.layout(ltr = True)
for node1, node2 in tree.links:
    ax1.plot(
        (node1.x, node2.x),
        (node1.y, node2.y),
        c = 'k'
    )
for leaf in tree.leaves:
    ax1.text(
        leaf.x + .1, 
        leaf.y - .1, 
        leaf.name,
        fontsize = 18
    )

#right-to-left layout with square links
tree.layout(ltr = False)
for node1, node2 in tree.links:
    ax2.plot(
        (node1.x, node1.x),
        (node1.y, node2.y)
    )
    ax2.plot(
        (node1.x, node2.x),
        (node2.y, node2.y)
    )

#clean up plots
ax1.set_xlim((-.5, 3.5))
ax2.set_xlim((-3.5, .5))
for ax in (ax1,ax2):
    ax.set_xticks([],[])
    ax.set_yticks([],[])

Neighbor joining tree construction and bootstrapping example:

from picea.algorithms import bootstrap as bts
import numpy as np
from sklearn.metrics import pairwise_distances

data = np.random.random_sample(size=(10, 5))
distance_matrix = pairwise_distances(data, metric="euclidean")
nj_tree, other_nj_trees = bts.prepare_bootstrap_trees_nj(distance_matrix, iteration=100)
nj_tree.bootstrap(other_nj_trees)

for n in nj_tree.breadth_first():
    if n.children is not None:
        print(f"node name: {n.name}, bootstrap value: {round(n.weight, 2)}%")

Produces following output:

node name: root, bootstrap value: 100.0%
node name: 16, bootstrap value: 21.0%
node name: 15, bootstrap value: 16.0%
node name: 14, bootstrap value: 48.0%
node name: 11, bootstrap value: 66.0%
node name: 13, bootstrap value: 47.0%
node name: 10, bootstrap value: 56.0%
node name: 12, bootstrap value: 43.0%

akdel/picea

picea