Spatially Resolved Transcriptomics Datasets Measured Utilizing Different Platforms.

In this tutorial, we demonstrate how to apply BatchEval to assess integrated data that is measured bt different platforms. As an example, we used three mouse olfactory bulb datas. One tissue section was profiled by 10x Genomics Visium, while the others were measured by Stereo-seq.

github: https://github.com/STOmics/BatchEval.git

Import packages

import os
import sys

sys.path.append(os.getcwd())

import scanpy as sc
from anndata import AnnData
from warnings import filterwarnings
from BatchEval import batch_eval
filterwarnings("ignore")

Load datasets

data = [sc.read_h5ad(os.path.join("./demo_data", t)) for t in sorted(os.listdir("./demo_data")) if t.endswith("h5ad")]

Optional: External data integration methods

import scanorama

merge_data = AnnData.concatenate(*data)

scanorama_correct = scanorama.correct_scanpy(
    [merge_data[merge_data.obs["batch"] == c] for c in merge_data.obs["batch"].cat.categories],
    return_dimred=True)

Found 22891 genes among all datasets
[[0.         0.50985222 0.        ]
 [0.         0.         0.02111486]
 [0.         0.         0.        ]]
Processing datasets (0, 1)

scanorama_correct = AnnData.concatenate(*scanorama_correct)

Running BatchEval Pipeline

batch_eval(*data,
           norm_log=True,
           is_scale=False,
           n_pcs=50,
           n_neighbors=15,
           batch_key="batch",
           position_key="X_umap",
           condition=None,
           count_key="total_counts",
           celltype_key="celltype",
           external_list=[{
               "correct_data": scanorama_correct,
               "re_pca": False,
               "re_neigh": True,
               "use_rep": "X_scanorama",
               "re_umap": True,
               "batch_key": "batch"},],
           report_path="./output/mouse_ob")

2023-09-18 16:00:18 BatchEval Pipeline Starting

2023-09-18 16:00:32.389107: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-09-18 16:00:37.907511: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT

2023-09-18 16:05:57 The 'Raw Report.html' has been saved to ./output/mouse_ob

Ouput

A series of HTML pages were output for the data integration evaluations, which we are able to find in thereport_path.

• BatchEval Report.html: the summaries report of BatchEval Pipeline.

• Raw Report.html: the raw data integration report.

• Harmony Report.html: the data integration report of using BatchEval Pipeline built-in Harmony mehtod, which is single cell base methods.

• BBKNN Report.html: the data integration report of using BatchEval Pipeline built-in BBKNN mehtod, which is single cell base methods.

• spatiAlign Report.html: the data integration report of using BatchEval Pipeline built-in spatiAlign mehtod, which is spatially resolved transcriptomics method.

• External Report.html: if the externnal_list is not None, and will generate external report.