analysis.py

import glob

import numpy as np
import pandas as pd
import scanpy as sc
from tabulate import tabulate
from tqdm import tqdm

# 获取当前目录下所有h5ad文件
h5ad_files = glob.glob("*.h5ad") + glob.glob("*.h5")

# 创建一个空的列表来存储结果
results = []

# 遍历每个h5ad文件
for file in tqdm(h5ad_files):
    try:
        # 读取h5ad文件
        adata = sc.read_h5ad(file)

        # 收集所需信息
        result = {
            "File": file,
            "X_max": float(adata.X.max()),
            "is_raw": np.max(np.abs(adata.X[:10] - adata.X[:10].astype(int))) < 1e-6,
            "has_raw": adata.raw is not None,
            "raw_X_max": float(adata.raw.X.max()) if adata.raw is not None else None,
            "obs_columns": ", ".join(sorted(list(adata.obs.columns))),
            "var_columns": ", ".join(sorted(list(adata.var.columns))),
            "Sample_obs_names": ", ".join(sorted(list(adata.obs_names))[:5]),
            "Sample_var_names": ", ".join(sorted(list(adata.var_names))[:5]),
            "N_cells": adata.n_obs,
            "N_genes": adata.n_vars,
        }

        results.append(result)

        # 关闭文件
        del adata

    except Exception as e:
        print(f"Error processing {file}: {str(e)}")

# 创建DataFrame
df = pd.DataFrame(results)
# sort df by file name
df.sort_values(by="File", inplace=True)

# 使用tabulate打印漂亮的表格
print(tabulate(df, headers="keys", tablefmt="grid", showindex=False))
df.to_csv("analysis.csv", index=False)