Question

我有一个数据框，其中包含（噬菌体）病毒样基因组之间的所有配对距离。

distData = pd.read_csv("distances.tab",sep="\t")
print(distData.head())
           reference-ID              query-ID  distance  p-value shared-hashes
0  H055-6_SuMu_C01_Hinf  H055-6_SuMu_C01_Hinf  0.000000      0.0     1000/1000
1  H049-6_SuMu_C01_Hinf  H055-6_SuMu_C01_Hinf  0.002000      0.0      921/1000
2  H013-2_SuMu_C01_Hinf  H055-6_SuMu_C01_Hinf  0.010914      0.0      660/1000
3  H081-1_SuMu_C01_Hinf  H055-6_SuMu_C01_Hinf  0.040448      0.0      272/1000
4  H058-3_SuMu_C01_Hinf  H055-6_SuMu_C01_Hinf  0.040310      0.0      273/1000

我想使用熊猫和seaborn来复制距离为方阵 squareDistMatrix 的带有彩色叶子的簇图。

        #Create the heatmap for distances
        squareDistMatrix = pd.pivot_table(distData, values='distance', index=['query-ID'], columns='reference-ID')
        squareDistMatrix.head()

        reference-ID    A001-2_Mu_C10_Aact  A001-3_Aaphi23_C25_Aact     A005-1_B3_C13_Aact  A010-2_B3_C13_Aact  A011-1_B3_C13_Aact  
    query-ID                                                                                    
    A001-2_Mu_C10_Aact  0.000000    1.0     1.000000    0.136948    0.295981    
    A001-3_Aaphi23_C25_Aact     1.000000    0.0     1.000000    1.000000    1.000000 
    A005-1_B3_C13_Aact  1.000000    1.0     0.000000    0.052915    0.050764    
    A010-2_B3_C13_Aact  0.136948    1.0     0.052915    0.000000    0.005942        
    A011-1_B3_C13_Aact  0.295981    1.0     0.050764    0.005942    0.000000

我关注了以下链接：

Setting col_colors in seaborn clustermap from pandas

https://seaborn.pydata.org/examples/structured_heatmap.html

但是这些链接不是来聚类映射由保存数据的原始数据帧生成的数据透视表。另外，我想通过每行名称中的子字符串为距离矩阵的叶子着色。那就是所有带有SuMu中间名称的病毒名称都属于相同超集群，属于进化组，因此它们被分配为相同颜色。

我用另一个stackoverflow帖子解决了它：

Column colors in clustermap of Python seaborn give unexpected results

我只是将解决方案粘贴给其他感兴趣的用户：

#Function to get the super cluster name from the full name of the phage
#Input: phageFullName "H055-6_SuMu_C01_Hinf" (String)
#Output: superClusterName "SuMu" (String)
def getSuperClusterName(phageFullName):
    return phageFullName.split("_")[1]
uniqueSuperClusters = distData['superCluster'].unique()
numUniqueSuperClusters = len(uniqueSuperClusters)
print("Unique superclusters: " + str(numUniqueSuperClusters))
#create distinct colors one for each supercluster
superClusterPalette = sns.husl_palette(numUniqueSuperClusters, s=.45)

#dict -> uniqueSuperClusterName: uniqueColour
superClusters2ColoursDict = dict(zip(uniqueSuperClusters, superClusterPalette))

#create a list of the supercluster of each row of the square distance matrix
superClusters = distData.superCluster
squareMatrixRows = list(squareDistMatrix.index)
superClusterInSquareMatrix = [] 
for row in squareMatrixRows:
    superClusterInSquareMatrix.append(getSuperClusterName(row))

#create a Series of supercluster as found in the distance matrix
superClusterSeries = pd.Series(superClusterInSquareMatrix)
#map each supercluster to a colour using the dictionary
superClusterColours = pd.Series(superClusterSeries).map(superClusters2ColoursDict)

#clustermap with colors in the rows
sns.clustermap(squareDistMatrix, metric="correlation", method="single", cmap="RdBu_r", standard_scale=1, row_colors=superClusterColours.values, linewidths=.75, figsize=(13, 13))

使用seaborn

0 个答案: