Cross species data format

[Hrovatin]

Files prepared separately for cVAEs/scVI and scGLUE/Saturn

• One to one orthologues:
  • For cVAE/scVI both species are in the same data: combined_orthologuesHVG.h5ad which has gene names mapped to mouse gene names
  • For scGLUE/Saturn they are in separate files: combined-hsPart_orthologuesHVG.h5ad (for hs/human) and combined-mmPart_orthologuesHVG.h5ad (for mm/mouse). Gene name correspondence (var names are species specific; gene symbols) is in combined_orthologuesHVG_geneMapping.tsv
• Genes regardless of orthology:
  • Only applicable to scGLUE/Saturn, same format as described for one to one orthologues.

If some genes are not present in the provided Saturn embeddings I guess we can filter them out for Saturn. But most of the gene symbols should be present hopefully.