论文阅读记录

STAGATE

  • 标题：Deciphering spatial domains from spatially resolved transcriptomics with an adaptive graph attention auto-encoder
  • 作者：University of Chinese Academy of Sciences
  • 发表会议/期刊：Nature Communications
  • 年份：2022
  • 链接：STAGATE

主要内容简介

  • Motivation: 邻域相似性是预定义的，无法自适应学习。

方法与创新点

  • 总体框架：

    Generalization

  • 方法概述：

    1. Construction of SNN (Construction of cell type-aware SNN (optional))

    2. Graph attention auto-encoder

      • Encoder

      Generalization

      • Decoder (迭代和encoder相似)

      Generalization

      • graph attention layer

      Generalization

      -graph attention layer with cell type-aware SNN

      Generalization

      • Loss function

      Generalization

任务

  1. Clustering based on STAGATE embeddings (mclust || Louvain)
  2. Spatial trajectory inference
  3. Identifying differentially expressed genes
  4. Identification of 3D spatial domains using STAGATE