将Matplotlib图像插入 pandas 数据框 [英] Insert matplotlib images into a pandas dataframe

问题描述

目的:我目前正在与rdkit一起根据rdkit.Chem.Draw.SimilarityMaps为分子的结构着色.现在，我想使用matplotlib图片SimilarityMaps函数将其引入到pandas数据框中，并以html文件的形式导出该表.

PURPOSE: I am currently working with rdkit to colour the structures of my molecules according to rdkit.Chem.Draw.SimilarityMaps. Now, I would like to use the matplotlib images SimilarityMaps function to introduce them in a pandas dataframe and export this table in the form of an html file.

代码:我尝试使用以下代码进行操作

CODE: I tried to do that with the following code

import pandas as pd
from rdkit import Chem
from rdkit.Chem import Draw
from rdkit.Chem.Draw import SimilarityMaps
from rdkit.Chem.Draw import IPythonConsole #Needed to show molecules
from rdkit.Chem.Draw.MolDrawing import MolDrawing, DrawingOptions

df = pd.DataFrame({'smiles':['Nc1nc(NC2CC2)c3ncn([C@@H]4C[C@H](CO)C=C4)c3n1','CCCC(=O)Nc1ccc(OCC(O)CNC(C)C)c(c1)C(C)=O','CCN(CC)CCNC(=O)C1=CC=C(C=C1)NC(=O)C','CC(=O)NC1=CC=C(C=C1)O','CC(=O)Nc1sc(nn1)[S](N)(=O)=O']})

def getSim(smi):
    mol = Chem.MolFromSmiles(smi)
    refmol = Chem.MolFromSmiles('c1ccccc1')
    fp = SimilarityMaps.GetMorganFingerprint(mol, fpType='bv')
    fig, maxweight = SimilarityMaps.GetSimilarityMapForFingerprint(refmol, mol, SimilarityMaps.GetMorganFingerprint)
    return fig

df['map'] = df['smiles'].map(getSim)
df.to_html('/.../test.html')

当我打开文件test.html时，地图列包含信息图形(200x200)".我检查我的数据框映射列是否包含对象:在python中可以，但在html文件中没有.

When I open the file test.html, the map column contains the information "Figure (200x200)". I check if my dataframe map column contains object: it's OK in python but not in html file.

问题::我不确定如何获取包含图像的数据框，我想在社区的帮助下阐明这一主题.

QUESTION: I'm not sure how to get a dataframe with images and I'd like to have the help of the community to clarify this subject.

预先感谢

推荐答案

您看到的Figure (200x200)是matplotlib Figure类的__repr__字符串.它是该python对象的文本表示形式(与执行print(fig)时所看到的相同).

What you see as Figure (200x200) is the __repr__ string of the matplotlib Figure class. It is the text representation of that python object (the same that you would see when doing print(fig)).

您想要的是在表中有一个实际的图像.一个简单的选择是将matplotlib图形另存为png图像，创建一个html标签<img src="some.png" />，从而显示该表.

What you want instead is to have an actual image in the table. An easy option would be to save the matplotlib figure as png image, create an html tag, <img src="some.png" /> and hence show the table.

import pandas as pd
import numpy as np;np.random.seed(1)
import matplotlib.pyplot as plt
import matplotlib.colors

df = pd.DataFrame({"info" : np.random.randint(0,10,10), 
                   "status" : np.random.randint(0,3,10)})

cmap = matplotlib.colors.ListedColormap(["crimson","orange","limegreen"])

def createFigure(i):
    fig, ax = plt.subplots(figsize=(.4,.4))
    fig.subplots_adjust(0,0,1,1)
    ax.axis("off")
    ax.axis([0,1,0,1])
    c = plt.Circle((.5,.5), .4, color=cmap(i))
    ax.add_patch(c)
    ax.text(.5,.5, str(i), ha="center", va="center")
    return fig

def mapping(i):
    fig = createFigure(i)
    fname = "data/map_{}.png".format(i)
    fig.savefig(fname)
    imgstr = '<img src="{}" /> '.format(fname)
    return imgstr


df['image'] = df['status'].map(mapping)
df.to_html('test.html', escape=False)

这样做的缺点是您在磁盘上的某处保存了很多图像.如果不希望这样做，则可以将编码为base64的图像存储在html文件<img src="data:image/png;base64,iVBORw0KGgoAAAAN..." />中.

The drawback of this is that you have a lot of images saved somewhere on disk. If this is not desired, you may store the image encoded as base64 in the html file, <img src="data:image/png;base64,iVBORw0KGgoAAAAN..." />.

import pandas as pd
import numpy as np;np.random.seed(1)
import matplotlib.pyplot as plt
import matplotlib.colors
from io import BytesIO
import base64

df = pd.DataFrame({"info" : np.random.randint(0,10,10), 
                   "status" : np.random.randint(0,3,10)})

cmap = matplotlib.colors.ListedColormap(["crimson","orange","limegreen"])

def createFigure(i):
    fig, ax = plt.subplots(figsize=(.4,.4))
    fig.subplots_adjust(0,0,1,1)
    ax.axis("off")
    ax.axis([0,1,0,1])
    c = plt.Circle((.5,.5), .4, color=cmap(i))
    ax.add_patch(c)
    ax.text(.5,.5, str(i), ha="center", va="center")
    return fig

def fig2inlinehtml(fig,i):
    figfile = BytesIO()
    fig.savefig(figfile, format='png')
    figfile.seek(0) 
    # for python 2.7:
    #figdata_png = base64.b64encode(figfile.getvalue())
    # for python 3.x:
    figdata_png = base64.b64encode(figfile.getvalue()).decode()
    imgstr = '<img src="data:image/png;base64,{}" />'.format(figdata_png)
    return imgstr

def mapping(i):
    fig = createFigure(i)
    return fig2inlinehtml(fig,i)


with pd.option_context('display.max_colwidth', -1):
    df.to_html('test.html', escape=False, formatters=dict(status=mapping))

输出看起来相同，但是没有图像保存到磁盘.

The output looks the same, but there are no images saved to disk.

在Jupyter笔记本中进行少量修改后，它也可以很好地工作

This also works nicely in a Jupyter Notebook, with a small modification,

from IPython.display import HTML
# ...
pd.set_option('display.max_colwidth', -1)
HTML(df.to_html(escape=False, formatters=dict(status=mapping)))

这篇关于将Matplotlib图像插入 pandas 数据框的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！