Installation and Usage

Installation

We developed the ARIEL-SRT package based on Python 3.9. ARIEL-SRT now is available on PyPI. To install it, run this command in the terminal:

pip install ariel-srt

The full package dependencies of ARIEL can be found in requirements.txt.

Usage

For data preprocessing, We first normalizes the expression for spots, then obtain embeddings using Profast or PCA to achieve data dimensionality reduction and eliminate potential batch effects. Specifically, when the samples are sourced from different slices, Profast is employed to eliminate batch effects(check data_processing/across_omics, data_processing/DLPFC, data_processing/hippo and the Visium dataset processing in data processing/hepatic lobule for details); while PCA is used when the samples are sourced from the same slices(check data_processing/hepatic lobule/Cosmx preprocessing for details)

We provide four examples to demonstrate the application of ARIEL:

DLPFC(Alignment)

In this Demo, we demonstrate the automatically generated landmarks by ARIEL under varying landmark quantities and their corresponding alignment results. Additionally, we apply spatial transformations such as rotation and mirroring to the slices to showcase the stability of the ARIEL.

Hippocampus

In this Demo, we use a Visium slice as the reference and integrate two Slide-SeqV2 slices with it via ARIEL.

Specifically, we use ARIEL to identify landmarks between two Slide-SeqV2 samples and the reference sample, and perform spatial alignment using these landmarks. We then employ Gaussian processes to transfer the expression information from Slide-SeqV2 sample to the reference.

Based on this approach, we achieve multi-sample information integration and generate several meaningful outputs. For instance, we constructe spatial gene expression maps based on the average gene information in the reference across the three samples, reflecting the common spatial expression characteristics of genes shared by all samples. Additionally, we achieve favorable spot clustering results using the averaged spatial expression data.

Brain

In this Demo, we integrate two adjacent brain slices from RIBOmap and STARmap technological platforms using ARIEL.

We use the RIBOmap sample as the reference and perform landmark identification, spatial alignment, and information transfer between the two slices through ARIEL. We use the transferred gene (information from the STARmap sample) to predict the protein (information from the RIBOmap sample). Compared to the baseline using single information source, incorporating gene information with ARIEL improves the accuracy of protein prediction for both randomly missing situation and regionally missing situation.

Hepatic lobule

In this Demo, we first integrate information from 10 Cosmx hepatic lobules to create landmark library and a spatial gene expression atlas of the lobules. We then introduce two Visium liver slices with different health states and use ARIEL to assist in lobule localization and analysis. By performing correlation analysis with the Cosmx atlas, we identify genes potentially associated with fatty liver disease.