用标签对纵向数据进行聚类？

Question

我的纵向数据如下：

import pandas as pd

# Define the updated data with samples only in 'sample_A' or 'sample_B'
data = {
    'gene_id': ['gene_1', 'gene_1', 'gene_1', 'gene_1', 'gene_1',
                'gene_1', 'gene_1', 'gene_1', 'gene_1', 'gene_1',
                'gene_2', 'gene_2', 'gene_2', 'gene_2', 'gene_2',
                'gene_2', 'gene_2', 'gene_2', 'gene_2', 'gene_2',
                'gene_3', 'gene_3', 'gene_3', 'gene_3', 'gene_3',
                'gene_3', 'gene_3', 'gene_3', 'gene_3', 'gene_3'],
    'position': [1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5],
    'value': [5.1, 5.5, 5.7, 6.0, 6.3,
              6.3, 6.5, 6.7, 6.8, 5.1,
              2.3, 2.5, 2.7, 3.0, 3.1,
              3.1, 3.2, 3.3, 3.4, 2.3,
              3.7, 3.8, 3.9, 4.0, 4.0,
              4.0, 4.1, 4.2, 4.3, 3.7],
    'sample': ['sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A',
               'sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A',
               'sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A']
}

# Create the DataFrame
df = pd.DataFrame(data)

我的目标是对基因值概况进行聚类，然后查看这些聚类如何与样本相对应。例如，这里的配置文件定义如下：获取样本，获取gene_id，现在获取结果子集中的所有（位置，值）元组。

通过在这里进行聚类，我有兴趣了解轮廓绘制的曲线的形状和幅度如何聚类。作为开始，简单的 KMeans 就适合我了。

聚类后，我们的想法是将其来自的样本恢复到每个配置文件，然后绘制聚类空间并查看样本如何分布。

我已经在 R 中看到过这个问题的解决方案，但在 python 中还没有看到任何解决方案。如有任何帮助，我们将不胜感激。

Answer 1

如果我正确理解您的问题，您可以使用下面的代码解决您的问题。基本上，您需要通过在

sample

和

gene_id

列上旋转数据框来重塑数据。每个配置文件将包含该样本中该基因的位置值。然后，您对数据应用 K 均值聚类（我根据基因数量使用了 3 个聚类，但您可以轻松更改）。我使用 PCA 来分解聚类并能够绘制数据。此外，我还计算了每个簇的样本分布：

import pandas as pd
from sklearn.cluster import KMeans
import matplotlib.pyplot as plt
from sklearn.decomposition import PCA

data = {
    'gene_id': ['gene_1', 'gene_1', 'gene_1', 'gene_1', 'gene_1',
                'gene_1', 'gene_1', 'gene_1', 'gene_1', 'gene_1',
                'gene_2', 'gene_2', 'gene_2', 'gene_2', 'gene_2',
                'gene_2', 'gene_2', 'gene_2', 'gene_2', 'gene_2',
                'gene_3', 'gene_3', 'gene_3', 'gene_3', 'gene_3',
                'gene_3', 'gene_3', 'gene_3', 'gene_3', 'gene_3'],
    'position': [1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5,
                 1, 2, 3, 4, 5],
    'value': [5.1, 5.5, 5.7, 6.0, 6.3,
              6.3, 6.5, 6.7, 6.8, 5.1,
              2.3, 2.5, 2.7, 3.0, 3.1,
              3.1, 3.2, 3.3, 3.4, 2.3,
              3.7, 3.8, 3.9, 4.0, 4.0,
              4.0, 4.1, 4.2, 4.3, 3.7],
    'sample': ['sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A',
               'sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A',
               'sample_A', 'sample_A', 'sample_A', 'sample_A', 'sample_B',
               'sample_B', 'sample_B', 'sample_B', 'sample_B', 'sample_A']
}

df = pd.DataFrame(data)

pivot_df = df.pivot_table(index=['sample', 'gene_id'], columns='position', values='value').reset_index()

# Perform K-means clustering
n_clusters = 3
kmeans = KMeans(n_clusters=n_clusters, random_state=42)

profiles = pivot_df.drop(columns=['sample', 'gene_id'])
kmeans.fit(profiles)

pivot_df['cluster'] = kmeans.labels_

# Reduce dimensions using PCA
pca = PCA(n_components=2)
profiles_pca = pca.fit_transform(profiles)

plot_df = pd.DataFrame(profiles_pca, columns=['PC1', 'PC2'])
plot_df['cluster'] = kmeans.labels_
plot_df['sample'] = pivot_df['sample']
plot_df['gene_id'] = pivot_df['gene_id']

# Build the plot
plt.figure(figsize=(8,6))
markers = {'sample_A': 'o', 'sample_B': 's'}
for cluster in range(n_clusters):
    cluster_data = plot_df[plot_df['cluster'] == cluster]
    
    for sample in ['sample_A', 'sample_B']:
        sample_data = cluster_data[cluster_data['sample'] == sample]
        plt.scatter(sample_data['PC1'], sample_data['PC2'], 
                    marker=markers[sample],
                    color=colors[cluster],
                    label=f'Cluster {cluster} - {sample}', 
                    alpha=0.7)

plt.title('KMeans Clustering of Gene Profiles with Sample Distribution (PCA Reduced)')
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.legend()
plt.show()

sample_cluster_distribution = pivot_df.groupby(['cluster', 'sample']).size().unstack().fillna(0)
print(sample_cluster_distribution)

我希望这是你想做的。 干杯！