PPubMed23 Dec 2024Cited 1×

STCGAN: a novel cycle-consistent generative adversarial network for spatial transcriptomics cellular deconvolution

Bo Wang, Yahui Long, Yuting Bai, Jiawei Luo, Chee Keong Kwoh

Abstract

The functions of complex tissues are intricately related to the spatial distribution of different cell types []. The development of spatial transcriptomics (ST) has revolutionized our ability to map gene expression patterns in native tissue contexts, providing unprecedented insights into cellular heterogeneity and spatial tissue . However, initial ST techniques lacked the resolution and gene coverage of single-cell RNA sequencing (scRNA-seq). For example, the popular 10x Visium platform can capture transcriptomes at scRNA-seq scale; however, it employs 55 \documentclass[12pt]{minimal} \usepack

§ The Valyu brief

The paper, in five minutes

Read from the full text via Valyu Answer

STCGAN uses a cycle-consistent GAN with a trainable cell‑to‑spot matrix and spatial regularization to outperform existing tools in spatial transcriptomics deconvolution.

What’s new

i.Introduced STCGAN, a cycle‑consistent GAN framework for spatial transcriptomics cellular deconvolution.
ii.Developed a trainable cell‑to‑spot mapping matrix to transfer scRNA‑seq annotations into spatial spots.
iii.Added spatial‑aware regularization to improve accuracy and robustness across datasets.
iv.Demonstrated superior performance over seven state‑of‑the‑art methods on five simulated and real ST datasets, including DLPFC and breast cancer samples.

How they did it

STCGAN first trains a cycle‑consistent adversarial network to model complex spatial gene expression patterns. It then learns a trainable cell‑to‑spot mapping matrix that projects scRNA‑seq reference data into the spatial transcriptomics space, while spatial‑aware regularization enforces smoothness across neighboring spots.

What they found

Outperformed seven benchmark deconvolution methods on five simulated and real datasets.
Showed robustness across varying gene richness and spatial resolutions.
Accurately captured cortical laminar structure in the DLPFC dataset and fine‑grained cell‑type heterogeneity in a human breast cancer case study.

Caveats

✕Relies on predefined cellular reference information, limiting applicability to unknown cell types.
✕Performance may degrade when reference datasets lack representation of certain tissue states.

§ The Valyu brief

Reading the full paper and taking notes. This takes a few seconds…

§ Ask this paper

Ask a question about this paper

Valyu reads the full text and answers from what the paper actually says.

§ Related work in other archives

Similar papers in all four archives

The same idea searched in arXiv, PubMed, bioRxiv and chemRxiv at once, so a method can lead you to where it is being applied.

AarXiv4 found

Searching the other archives…