#!/usr/bin/env python
# coding: utf-8

# In[1]:


import celltypist
from celltypist import models
import scanpy as sc
import pandas as pd 
import numpy as np
import anndata
import re
import h5py
import scipy.sparse as scs
import concurrent.futures


# In[2]:


def read_mat(h5_con):
    mat = scs.csc_matrix(
        (h5_con['matrix']['data'][:], # Count values
         h5_con['matrix']['indices'][:], # Row indices
         h5_con['matrix']['indptr'][:]), # Pointers for column positions
        shape = tuple(h5_con['matrix']['shape'][:]) # Matrix dimensions
    )
    return mat

# define a function to obeservation (i.e. metadata)

def read_obs(h5con):
    bc = h5con['matrix']['barcodes'][:]
    bc = [x.decode('UTF-8') for x in bc]

    # Initialized the DataFrame with cell barcodes
    obs_df = pd.DataFrame({ 'barcodes' : bc })

    # Get the list of available metadata columns
    obs_columns = h5con['matrix']['observations'].keys()

    # For each column
    for col in obs_columns:
        # Read the values
        values = h5con['matrix']['observations'][col][:]
        # Check for byte storage
        if(isinstance(values[0], (bytes, bytearray))):
            # Decode byte strings
            values = [x.decode('UTF-8') for x in values]
        # Add column to the DataFrame
        obs_df[col] = values
    
    return obs_df
# define a function to construct anndata object from a h5 file
def read_h5_anndata(h5_file):
    h5_con = h5py.File(h5_file, mode = 'r')
    # extract the expression matrix
    mat = read_mat(h5_con)
    # extract gene names
    genes = h5_con['matrix']['features']['name'][:]
    genes = [x.decode('UTF-8') for x in genes]
    # extract metadata
    obs_df = read_obs(h5_con)
    # construct anndata
    adata = anndata.AnnData(mat.T,
                             obs = obs_df)
    # make sure the gene names aligned
    adata.var_names = genes

    adata.var_names_make_unique()
    return adata


# In[3]:


meta_data=pd.read_csv("hise_meta_data_2023-11-19.csv")


# In[4]:


results = []
for file_name in meta_data['file.path']:
    result = read_h5_anndata(file_name)
    results.append(result)
adata = anndata.concat(results)


# In[5]:


adata.write_h5ad('adata_all_raw.h5ad')


# In[ ]: