Reproduce “SPHENIC: Topology-Aware Multi-View Clustering for Spatial Transcriptomics” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Chenkai Guo, Yikai Zhu, Renxiang Guan, Jinli Ma, Siwei Wang, Ke Liang, Guangdun Peng, Dayu Hu

SPHENIC introduces a topology-aware multi-view clustering framework for spatial transcriptomics data, enhancing the accuracy of spatial domain identification by integrating topological invariants and enforcing spatial coherence. The system demonstrated superior performance, achieving up to 9.14% higher Adjusted Rand Index compared to previous state-of-the-art methods on benchmark datasets.

Key insights

  • Extended Persistent Homology can effectively capture global topological features in spatial transcriptomics, significantly improving the robustness of cell representations against noise.
  • Explicitly enforcing spatial coherence via a dedicated loss function during embedding learning is critical for ensuring that physically adjacent cells remain close in the latent space, producing more biologically meaningful clusters.
  • Integrating a ZINB probabilistic reconstruction loss adeptly models the sparse and overdispersed nature of spatial transcriptomics data, contributing to enhanced accuracy in gene expression recovery and clustering.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.