Tutorials

An scglue workflow can be divided into the following two stages. We provide separate tutorials for each stage, using the integration of scRNA-seq and scATAC-seq data as an example.

• Stage 1: Data preprocessing
  • Read data
  • Preprocess scRNA-seq data
  • Preprocess scATAC-seq data
  • Construct prior regulatory graph
    • Obtain genomic coordinates
    • Graph construction
  • Save preprocessed data files
• Stage 2: Model training
  • Read preprocessed data
  • Configure data
  • Train GLUE model
  • Check integration diagnostics
  • Apply model for cell and feature embedding
• Stage 3: Regulatory inference
  • Read intermediate results
  • Cis-regulatory inference with GLUE feature embeddings
  • Visualize the inferred cis-regulatory regions
  • Construct TF-gene regulatory network from inferred cis-regulatory regions
    • Draft a coexpression-based network
    • Generate TF cis-regulatory ranking bridged by ATAC peaks
    • Generate TF cis-regulatory ranking with proximal promoters (optional)
    • Prune coexpression network using cis-regulatory ranking
  • Visualize the inferred TF-target gene network
• Extra: Model training on partially paired data
  • Read preprocessed data
  • Label paired cells via obs_names
  • Configure data
  • Train GLUE model
  • Continue as in stage 2