查看原文
其他

ETE构建、绘制进化树

2017-07-21 陈同 生信宝典

ETE能做什么

A Python framework for construction, analysis and visualization of trees.

安装和使用

  • conda安装

    # Install Minconda  (you can ignore this step if you already have Anaconda/Miniconda)
    wget http://repo.continuum.io/miniconda/Miniconda-latest-Linux-x86_64.sh -O Miniconda-latest-Linux-x86_64.sh
    bash Miniconda-latest-Linux-x86_64.sh -b -p ~/anaconda/
    export PATH=~/anaconda/bin:$PATH;

    # Install ETE
    conda install -c etetoolkit ete3 ete3_external_apps

    # Check installation
    ete3 version
    ete3 build check
  • github源码安装

    wget https://github.com/etetoolkit/ete/archive/master.zip -O ete3.20160719.zip
    unzip ete3.20160719.zip
    python setup.py install
    yum install python-six.noarch
    ete3 upgrade-external-tools

ETE运行

  • 输入序列

    • 输入序列为标准的fasta格式文件,名字可以为任意形式

    • 如果需要在序列名字中区分物种信息,fasta序列名需满足SpeciesCode_SequenceName例如HUMAN_p53 = HUMAN, p53。 可以通过参数--spname-delimiter指定使用其它字符作为分隔符。

  • ete预先定义了多个流程用以完成从原始fasta序列到后续进化树生成的各个步骤。

    运行以下命令可以列出系统自带的流程及其解释

    ete3 build workflows genetree
  • 使用既定流程最简单运行

    -w指定所用的流程,-a指定输入序列,--tools-dir指定安装的外部程序的路径

    ete3 build -w standard_fasttree -a diTPS.prot.fa -o standard_fasttree
         --tools-dir /root/.etetoolkit/ext_apps-latest/
  • 自己定制流程

    • 查看已有分析模块的定义

      ete3 build show phyml_default

      [phyml_default]
        _desc = Phyml tree using +G+I+F, 4 classes and aLRT branch supports. Default models JTT/GTR
        _app = phyml
        _aa_model = JTT
        -nt_model = GTR
           --pinv = e
          --alpha = e
       --nclasses = 4
               -o = tlr
               -f = m
      --bootstrap = -2
    • 修改部分定义获得新的模块

      ete3 build show phyml_default >customized.config

      ## 修改后的customized.config

      [phyml_bootstrap_100]
             _desc = Phyml tree using +G+I+F, 4 classes and aLRT branch supports. Default models JTT/GTR
              _app = phyml
         _aa_model = JTT
         -nt_model = GTR
            --pinv = e
           --alpha = e
        --nclasses = 4
                -o = tlr
                -f = m
       --bootstrap = 100
      [trimal_auto]
                _desc = trimal alignment cleaning using auto algorithm
              _app  = trimal
       -automated1 =

      ## 使用新定义的模块
      ete3 build -a diTPS.prot.fa --clearall -o phyml_bootstrap_100 -w
      mafft_einsi-trimal_auto-none-phyml_bootstrap_100 -c customized.cfg --cpu 5
    • totally 4 parts included as stated above, multiple sequence alignment, trimming MSA results, select best model, use appropriate softwares to build tree.

    • - represents command separator

    • none represents skipping related operations

    • 获取可以定制的各部分命令

      ete3 build apps
    • 选择预定义好的模块,如tree builders: phyml_default_bootstrap, aligners: mafft_einsi, model testers: pmodeltest_full_slow, alg cleaners:trimal_gappyout.

    • 流程定制模板: 顺序为aligner-trimmer-model_tester-builder

    • 基于我们的选择定制的流程mafft_einsi-trimal_gappyout-pmodeltest_full_slow-phyml_default_bootstrap

      ete3 build -w mafft_einsi-trimal_gappyout-pmodeltest_full_slow-phyml_default_bootstrap
        -a diTPS.prot.fa -o custom_phymltree
    • -w可以接受多个流程(空格分开),进而得到不同的比对工具、处理方式和建树工具 输出的多个结果,可以通过ete3 compare比较这些结果的吻合度, 比如Robinson-Foulds距离等。

      ete3 compare -r newtree1.nwq -t "tree2.nw tree3.nw tree4.nw" --unrooted

      # Tree file can be got using find
      find custom_phymltree -name *.nw
    • 定制不同的分析模块

  • 氨基酸比对指导核苷酸比对的进化树构建 (要求氨基酸序列与核苷酸序列名字一一对应,核苷酸序列可以含有终止密码子,最终获得的核苷酸比对序列存储在*.used_alg.fa文件中。)

    ete3 build -a diTPS.prot.fa -n diTPS.nucl.fa -o aa2nt
      -w standard_fasttree --clearall --nt-switch-threshold 0.9
      -C 20
  • 使用预先比对好的序列, 使用none代替aligner

    ete3 build -a diTPS.prot.aln.fa -w none-none-none-fasttree
      -o manual_alg --clearall
  • 设置树的根节点

    from ete3 import Tree

    tree = Tree('tree.nw')

    root = 'one_node_name'
    tree.set_outgroup(root)

    #use mid-point as root
    mid = tree.get_midpoint_outgroup()
    tree.set_outgroup(mid)

    tree.write('tree.rooted.nw')
    tree.render('tree.rooted.pdf')

问题解决

  • ETE: cannot connect to X server 如果程序运行出现错误ETE: cannot connect to X server则安装Xvfb, 并运行 xvfb-run ete3取代ete3, 后面的代码不变。

    yum install xorg-x11-server-Xvfb.x86_64
    xvfb-run ete3 build -w standard_fasttree -a diTPS.prot.fa -o standard_fasttree
  • ETE: cannot connect to X server (solve in python script or jupyter ref)

    # Add the following 4 lines at the beginning of python code
    # or the first cell in Jupyter
    from xvfbwrapper import Xvfb

    vdisplay = Xvfb()
    vdisplay.start()

    # launch stuff inside virtual display here
    # other python codes here

    # Add this line at the end of python code
    # or the last cell in Jupyter
    vdisplay.stop()
    • Install xvfbwrapper using pip install xvfbwrapper    

  • External applications directory are not found 指定ETE使用的工具的安装路径;一般发生在普通用户使用根用户编译的ETE时。

    --tools-dir /root/.etetoolkit/ext_apps-latest/

Tree annotation

# A virtual X-server XVFB is used in case you do not have X-server
from xvfbwrapper import Xvfb

vdisplay = Xvfb()
vdisplay.start()

# launch stuff inside virtual display here

#vdisplay.stop()

from ete3 import Tree, faces, TreeStyle, NodeStyle
from ete3 import ClusterTree, RectFace, AttrFace, ProfileFace, TextFace
from ete3.treeview.faces import add_face_to_node
import pandas as pd
import numpy as np
import colorsys

The most simple way of showing a tree.

t = Tree()
t.populate(7,names_library=['A','B','C','D','E','F','G'])
## %%liline is used for showing plots in ipythonnotebook.
## t.render(file_name="tree.pdf") # will save tree into pdf file
t.render(file_name="%%inline")

Get the randomly generated tree in newick format and save to a string variable which can be read using Tree() function.

t_str = t.write(outfile=None, format=0)
t_str

'(((B:1,A:1)1:1,(G:1,F:1)1:1)1:1,(E:1,(D:1,C:1)1:1)1:1);'
t = Tree(t_str)
ts = TreeStyle()
ts.show_leaf_name = True
ts.show_branch_length = True
ts.show_branch_support = True
t.render(file_name="%%inline", tree_style=ts)

Get the randomly generated tree in newick format and save to file which can also be read using Tree() function.

t.write(outfile="tree.nw", format=0)
t = Tree("tree.nw")
ts.mode = "c"
ts.arc_start = -180 # 0 degrees = 3 o'clock
ts.arc_span = 180
t.render(file_name="%%inline", w=500, tree_style=ts)

设置根节点、叶节点和中间节点的属性

ts = TreeStyle()
ts.show_leaf_name = True
ts.show_branch_length = True
ts.show_branch_support = True

# Draws nodes as small red spheres of diameter equal to 10 pixels
for n in t.traverse():  # Traverse each node and set attribute for each type of nodes
   if n.is_leaf(): # Decide if leaf node
       nstyle = NodeStyle()
       nstyle["shape"] = "sphere"
       nstyle["size"] = 10
       nstyle["fgcolor"] = "darkred"
       n.set_style(nstyle)
   else:
       nstyle = NodeStyle()
       nstyle["shape"] = "square"
       nstyle["size"] = 15
       nstyle["fgcolor"] = "orange"
       n.set_style(nstyle)        

t.img_style["size"] = 30
t.img_style["fgcolor"] = "blue"

t.render(file_name="%%inline", w=500, tree_style=ts)

修改节点的名字

t = Tree(t_str)

nameMap = {'A': 'American', 'B': 'Britain', 'C':'China',
          'D':'Dutch', 'E':'Egypt','F':'France','G':'German'}

ts = TreeStyle()
ts.show_leaf_name = False
ts.show_branch_length = True
ts.show_branch_support = True

# Creates my own layout function. I will use all previously created
# faces and will set different node styles depending on the type of
# node.
def mylayout(node):  
   # If node is a leaf, add the nodes name and a its scientific
   # name
   if node.is_leaf():
       # We can also create faces on the fly
       newName = nameMap.get(node.name, node.name)
       newNameFace = faces.TextFace(newName)
       faces.add_face_to_node(newNameFace, node, column=1, aligned=True)

       # Sets the style of leaf nodes
       node.img_style["size"] = 12
       node.img_style["shape"] = "sphere"
       node.img_style["fgcolor"] = "blue"
   #If node is an internal node
   else:
       # Sets the style of internal nodes
       node.img_style["size"] = 1
       node.img_style["shape"] = "circle"
       node.img_style["fgcolor"] = "darkred"

ts.layout_fn = mylayout

t.img_style["size"] = 30
t.img_style["fgcolor"] = "black"

t.render(file_name="%%inline", w=600, tree_style=ts)

不同的节点设置不同的背景色,树枝的颜色

t = Tree(t_str)

nameMap = {'A': 'American', 'B': 'Britain', 'C':'China',
          'D':'Dutch', 'E':'Egypt','F':'France','G':'German'}

colorMap = { 'American': '#ACFFFF',
            'Britain': '#ACACFF',
            'China': '#ACACAC',
            'Dutch': '#59ACAC',
            'Egypt': '#5959AC',
            'France': '#595959',
            'German': '#065959'}

ts = TreeStyle()
ts.show_leaf_name = False
ts.show_branch_length = True
ts.show_branch_support = True

# Creates my own layout function. I will use all previously created
# faces and will set different node styles depending on the type of
# node.
def mylayout(node):  
   #Change branch color
   node.img_style["hz_line_color"] = 'orange'  # change horizontal branch color
   node.img_style["vt_line_color"] = 'red' # Change vertical branch color

   # If node is a leaf, add the nodes name and a its scientific
   # name
   if node.is_leaf():
       # We can also create faces on the fly
       newName = nameMap.get(node.name, node.name)
       newNameFace = faces.TextFace(newName)
       faces.add_face_to_node(newNameFace, node, column=1, aligned=True)

       # Sets the style of leaf nodes
       node.img_style["size"] = 12
       node.img_style["shape"] = "sphere"
       node.img_style["fgcolor"] = "blue"
       node.img_style["bgcolor"] = colorMap[newName]

       node.img_style["hz_line_color"] = 'blue'  # change branch color
   #If node is an internal node
   else:
       # Sets the style of internal nodes
       node.img_style["size"] = 1
       node.img_style["shape"] = "circle"
       node.img_style["fgcolor"] = "darkred"

ts.layout_fn = mylayout

t.img_style["size"] = 30
t.img_style["fgcolor"] = "black"

t.render(file_name="%%inline",tree_style=ts)

树+热图(自定义颜色+列名字)

自定义热图函数,同时该函数也支持替换或新增节点的名字。

nameFace = AttrFace("name", fsize=12) #Set leaf node attribute

def setup_heatmap(tree, tree_style, header, center_value=0.0, nameMap ={}, nameLabel = '',
                 color_up=0.7, color_down=0.2, color_center="white"):

   DEFAULT_COLOR_SATURATION = 0.5
   BASE_LIGHTNESS = 0.7
   def gradient_color(value, max_value, saturation=0.5, hue=0.1):    
       def rgb2hex(rgb):
           return '#%02x%02x%02x' % rgb
       def hls2hex(h, l, s):
           return rgb2hex( tuple(map(lambda x: int(x*255),
                         colorsys.hls_to_rgb(h, l, s))))

       lightness = 1 - (value * BASE_LIGHTNESS) / max_value
       return hls2hex(hue, lightness, DEFAULT_COLOR_SATURATION)

   # Calculate max gradient value from the ClusterTree matrix
   maxv = abs(center_value - tree.arraytable._matrix_max)
   minv = abs(center_value - tree.arraytable._matrix_min)
   if center_value <= tree.arraytable._matrix_min:
       MAX_VALUE = minv + maxv
   else:
       MAX_VALUE = max(maxv, minv)

   # Add heatmap colors to tree
   cols_add_before_heat = 0
   if nameMap:
       cols_add_before_heat = 1
   for lf in tree:
       if nameMap:
           longNameFace = faces.TextFace(nameMap.get(lf.name, lf.name))
           lf.add_face(longNameFace, column=0, position="aligned")

       for i, value in enumerate(getattr(lf, "profile", [])):
           if value > center_value:
               color = gradient_color(abs(center_value - value), MAX_VALUE,
                                      hue=color_up)
           elif value < center_value:
               color = gradient_color(abs(center_value - value), MAX_VALUE,
                                      hue=color_down)
           else:
               color = color_center
           lf.add_face(RectFace(20, 20, color, color), position="aligned",
                       column=i+cols_add_before_heat)
           # Uncomment to add numeric values to the matrix
           #lf.add_face(TextFace("%0.2f "%value, fsize=5), position="aligned", column=i)
       lf.add_face(nameFace, column=i+cols_add_before_heat+1, position="aligned")

   if nameMap and nameLabel:
       nameF = TextFace(nameLabel, fsize=7)
       #nameF.rotation = -90
       tree_style.aligned_header.add_face(nameF, column=0)
   # Add header
   for i, name in enumerate(header):
       nameF = TextFace(name, fsize=7)
       nameF.rotation = -90
       tree_style.aligned_header.add_face(nameF, column=i+cols_add_before_heat)
#-------------END setup_heatmap----------------------------------------------

读入矩阵 (可把文后的测试矩阵存储到文件中读入)

矩阵需满足三个条件:

  • 矩阵为TAB键分割,第一行是每列的名字

  • 矩阵每一行第一列为行名字,与树的节点对应

  • 矩阵可以存储与一个文件中,也可以是如下的字符串

data = pd.read_table("matrix", header=0, index_col=0)
data.index.name = "#Names"  #修改第一行的名字使其符合ETE的要求
data_mat = data.to_csv(None, sep="\t", float_format="%.2f")
header = list(data.columns.values)  #获取列的名字用于标记

data

col1col2col3col4col5col6col7
#Names






A-1.23-0.811.790.78-0.42-0.690.58
B-1.76-0.941.160.360.41-0.351.12
C-2.190.130.65-0.510.521.040.36
D-1.22-0.980.79-0.76-0.291.540.93
E-1.47-0.830.850.07-0.811.530.65
F-1.04-1.110.87-0.14-0.801.740.48
G-1.57-1.171.290.23-0.201.170.26
data_mat'#Names\tcol1\tcol2\tcol3\tcol4\tcol5\tcol6\tcol7\nA\t-1.23\t-0.81\t1.79\t0.78\t-0.42\t-0.69\t0.58\nB\t-1.76\t-0.94\t1.16\t0.36\t0.41\t-0.35\t1.12\nC\t-2.19\t0.13\t0.65\t-0.51\t0.52\t1.04\t0.36\nD\t-1.22\t-0.98\t0.79\t-0.76\t-0.29\t1.54\t0.93\nE\t-1.47\t-0.83\t0.85\t0.07\t-0.81\t1.53\t0.65\nF\t-1.04\t-1.11\t0.87\t-0.14\t-0.80\t1.74\t0.48\nG\t-1.57\t-1.17\t1.29\t0.23\t-0.20\t1.17\t0.26\n'header['col1', 'col2', 'col3', 'col4', 'col5', 'col6', 'col7']

调用函数绘制热图

t = ClusterTree(t_str, data_mat)

ts = TreeStyle()
ts.show_leaf_name = False
ts.show_branch_length = True
ts.show_branch_support = True

setup_heatmap(t, ts, header, center_value=0, color_up=0.9, color_down=0.3, color_center="white")

t.render(file_name="%%inline", tree_style=ts)

绘制热图时修改Layout

def mylayout_only(node):  
   #Change branch color
   node.img_style["hz_line_color"] = 'orange'  # change horizontal branch color
   node.img_style["vt_line_color"] = 'red' # Change vertical branch color

   # If node is a leaf, add the nodes name and a its scientific
   # name
   if node.is_leaf():
       newName = nameMap.get(node.name)
       node.img_style["size"] = 12
       node.img_style["shape"] = "sphere"
       node.img_style["fgcolor"] = "blue"
       node.img_style["bgcolor"] = colorMap[newName]        
       node.img_style["hz_line_color"] = 'blue'  # change branch color
   #If node is an internal node
   else:
       # Sets the style of internal nodes
       node.img_style["size"] = 1
       node.img_style["shape"] = "circle"
       node.img_style["fgcolor"] = "darkred"

t = ClusterTree(t_str, data_mat)

ts = TreeStyle()
ts.show_leaf_name = False
ts.show_branch_length = True
ts.show_branch_support = True
ts.layout_fn = mylayout_only

setup_heatmap(t, ts, header, center_value=0, color_up=0.9, color_down=0.3,
             color_center="white", nameMap=nameMap, nameLabel="Full")

t.render(file_name="%%inline", tree_style=ts)

测试矩阵

## 矩阵为TAB键分割 ## 矩阵每一行第一列为行名字,与树的节点对应 ## 矩阵可以存储于一个文件中,也可以是如下的字符串 matrix = """ #Names\tcol1\tcol2\tcol3\tcol4\tcol5\tcol6\tcol7 A\t-1.23\t-0.81\t1.79\t0.78\t-0.42\t-0.69\t0.58 B\t-1.76\t-0.94\t1.16\t0.36\t0.41\t-0.35\t1.12 C\t-2.19\t0.13\t0.65\t-0.51\t0.52\t1.04\t0.36 D\t-1.22\t-0.98\t0.79\t-0.76\t-0.29\t1.54\t0.93 E\t-1.47\t-0.83\t0.85\t0.07\t-0.81\t1.53\t0.65 F\t-1.04\t-1.11\t0.87\t-0.14\t-0.80\t1.74\t0.48 G\t-1.57\t-1.17\t1.29\t0.23\t-0.20\t1.17\t0.26 """

产生颜色的辅助函数,给定一个列表,这个函数会自动返回一个字典包含每个字段对应的颜色。

def hex2rgb(hexcolor):    return [(hexcolor>>16) & 0xff, (hexcolor>>8) & 0xff, hexcolor & 0xff] def rgb2hex(rgbcolor):    r, g, b = rgbcolor    rgb = hex((r << 16) + (g << 8) +b)[2:].upper()    zero = '0'* (6-len(rgb))    return '#'+zero+rgb #---------------------------------- def generateColor(labelL):    labelL = list(set(labelL))    labelL.sort()    colorD = {}    r = 255    g = 255    b = 255    len_label = int(len(labelL) / 3 + 1)    step = int(250 / len_label)    cnt = 1    for labels in labelL:        if cnt % 3 == 1:            r = r - step        elif cnt % 3 == 2:            g = g -step        else:            b = b - step        cnt += 1        color = rgb2hex([r, g, b])        colorD[labels] = color    return colorD clan_colorD = generateColor(nameMap.values()) clan_colorD {'American': '#ACFFFF', 'Britain': '#ACACFF', 'China': '#ACACAC', 'Dutch': '#59ACAC', 'Egypt': '#5959AC', 'France': '#595959', 'German': '#065959'}

带有Support value的Newick树,供测试不同的属性使用

nw = """
(((Dre:0.008339,Dme:0.300613)1.000000:0.596401,
(Cfa:0.640858,Hsa:0.753230)1.000000:0.182035)1.000000:0.106234,
((Dre:0.271621,Cfa:0.046042)1.000000:0.953250,
(Hsa:0.061813,Mms:0.110769)1.000000:0.204419)1.000000:0.973467);

关闭Virtual X-server

vdisplay.stop()

Ipython notebook for easy usage

https://github.com/Tong-Chen/notebook/blob/master/ETE.ipynb

Reference

  • ETE tutorial http://etetoolkit.org/docs/latest/tutorial/index.html

  • ETE googlegroup https://groups.google.com/d/topic/etetoolkit/pXr4B71Ozt0

Blog link

原文链接 http://blog.genesino.com/2016/07/ete/

R绘图学习

R语言学习  -  入门环境Rstudio

R语言学习  - 热图绘制  (heatmap)

R语言学习  -  基础概念和矩阵操作

R语言学习  -  热图美化

R语言学习 - 热图简化

R语言学习 - 线图绘制

R语言学习 - 线图一步法

R语言学习 - 箱线图(小提琴图、抖动图、区域散点图)

R语言学习 - 箱线图一步法

R语言学习 - 富集分析泡泡图 (文末有彩蛋)

文章集锦

Linux学习  R统计绘图  Python教程  Perl学习

生信傻瓜  NGS持续更新中 

(链接失效,点击菜单知识库查看)

关注生信宝典,换个角度学生信

关注宏基因组,再专业一点


您可能也对以下帖子感兴趣

文章有问题?点此查看未经处理的缓存