Healthy Heart Dataset Processing Script

This script processes the Healthy Heart dataset obtained from Figshare
(Yu et al., 2023).

Dataset Location
The dataset should be saved in scMEDAL_for_scRNAseq/Experiments/data/HealthyHeart_data/raw.

Preprocessing and Output
The script preprocesses the data and stores the processed output in
scMEDAL_for_scRNAseq/Experiments/data/HealthyHeart_data/scenario_id.
By default, scenario_id is set to log_transformed_3000hvggenes.

Environment
This script should be run in the preprocess_and_plot_umaps_env environment.

import anndata

print(anndata.__version__)
import os

from scMEDAL.utils.preprocessing_utils import H5ADLoader, scRNAseq_pipeline_log, scRNAseq_pipeline_loghvg
from scMEDAL.utils.utils import save_adata
import sys

# To import data_base_path from paths_config
# Add the parent directory to the Python path
sys.path.append("../")

# Now you can import from the parent directory
from paths_config import data_base_path,scenario_id

# Path to the directory containing raw data
# Example: /MyscMEDALExpt/Experiments/data/HealthyHeart_data/raw
parent_path = os.path.join(data_base_path, "raw")
# file name
healthy_heart_file = "Healthy_human_heart_adata.h5ad"
file_path = os.path.join(parent_path,healthy_heart_file )


# Load adata with H5ADLoader
loader = H5ADLoader(file_path)
loader.load_h5ad()
adata = loader.create_anndata()
# save adata raw
# save_adata(adata,parent_path)

Reading X
CSR matrix created successfully with shape: [486134  33538], Transposed: False
Reading: X
Reading: layers
No data found under key 'layers'
Reading: obs
Reading: obsm
Reading: obsp
No data found under key 'obsp'
Reading: uns
Reading: var
Reading: varm
No data found under key 'varm'
Reading: varp
No data found under key 'varp'

adata

AnnData object with n_obs × n_vars = 486134 × 33538
    obs: 'Age', 'AgeBin', 'DeathType', 'DonorID', 'Gender', 'Organ', 'Race', 'SampleType', 'Source', 'Tissue', 'TissueDetail', '_index', 'celltype', 'protocol', 'sampleID'
    var: '_index', 'gene_ids'
    uns: 'celltype_colors'
    obsm: 'X_umap'

save_data = False
# expt = "expt_name"
expt = scenario_id # On this case the scenario_id has predefined preprocessing 
print("expt:",scenario_id)
if save_data:
    out_path = os.path.join(data_base_path,expt)
    print(out_path)
    # Check if the directory exists, if not, create it
    if not os.path.exists(out_path):
        os.makedirs(out_path)
# Adapt to the type of preprocessing you need
# Log transform log(x+1)
if expt == "log_transformed":
    adata_log = scRNAseq_pipeline_log(adata, min_genes_per_cell=10, min_cells_per_gene=3, total_counts_per_cell=10000)
    if save_data:
        save_adata(adata_log,out_path)

# Log transform log(x+1) +HVG genes
elif expt == "log_transformed_3000hvggenes":
    adata_log_hvg = scRNAseq_pipeline_loghvg(adata, min_genes_per_cell=10, min_cells_per_gene=3, total_counts_per_cell=10000,n_top_genes=3000)
    if save_data:
        save_adata(adata_log_hvg,out_path)

expt: log_transformed_3000hvggenes

adata_log_hvg

AnnData object with n_obs × n_vars = 486134 × 3000
    obs: 'Age', 'AgeBin', 'DeathType', 'DonorID', 'Gender', 'Organ', 'Race', 'SampleType', 'Source', 'Tissue', 'TissueDetail', '_index', 'celltype', 'protocol', 'sampleID', 'n_genes'
    var: '_index', 'gene_ids', 'n_cells', 'highly_variable', 'means', 'dispersions', 'dispersions_norm'
    uns: 'celltype_colors', 'log1p', 'hvg'
    obsm: 'X_umap'

adata_log_hvg.var