Visium kidney cancer dataset at the tumour-normal interface

The kidney cancer dataset profiled by 10x Visium and paired scRNA-seq data [LFL+22] are available to download as h5ad objects in Mendeley Data (https://doi.org/10.17632/g67bkbnhhg.1). We used a sample with the donorid ‘PD47171’ in our analysis.

• Cell type deconvolution using MacSGP
• Applying MacSGP to detect cell-type-specific SGPs
• Kidney cancer dataset analysis
  • Results of RCC
  • Spatial gradient analysis
• Kidney cancer: per-cell-type K selection (reproduce)
  • Part 1 — compute: synchronized per-cell-type K selection
  • Part 2 — compute: pathway enrichment of the selected SGPs (poles separated)
  • Part 3 — figures
    • 1. Cumulative held-out ΔLL vs K (K>0 cell types)
    • 2. Spatial distribution of all selected SGP factors
    • 3. Pathway enrichment of the SGPs
• Results of k=10 on the kidney cancer dataset
• Robustness analysis with random initilization
• Sensitivity Analysis — Factor & Loading Coherence
  • Section 1: coef_reg Sensitivity
  • Section 2: coef_fe Sensitivity
• Comparison with other baselines on the kidney cancer dataset
  • Comparison with cell-type-restricted analysis
  • Comparison with domain-based analysis
  • Comparison with factor methods

References

[LFL+22]

Ruoyan Li, John R Ferdinand, Kevin W Loudon, Georgina S Bowyer, Sean Laidlaw, Francesc Muyas, Lira Mamanova, Joana B Neves, Liam Bolt, Eirini S Fasouli, and others. Mapping single-cell transcriptomes in the intra-tumoral and associated territories of kidney cancer. Cancer cell, 40(12):1583–1599, 2022.