Human CD4 and CD8 TILs references

[gianfilippo]

HI,

I could not find a description of how you 'transformed' the original data from the Science paper into a ProjecTILs reference.

I am asking since in Fig 1 (Zhang et al.) the CD8 UMAP has 17 clusters grouped into 8 large categories (Tn, Tem, Temra, Tex, Trm,Tc17, ISG+, KIR+ NK-like). The reference, instead, has 7 categories (Tn, Tem, Temra, Tex, Tcm, Tpex, MAIT) and only the first 4 match.

Could you please clarify ?

Thanks

[mass-a]

Hello Gianfilippo,

we did not use the cell type annotations from the Zhang et al. Science paper. Their reference was integrated using harmony and cannot be directly used as a ProjecTILs reference, as it does not return corrected expression counts.

The CD4 T cell reference was indeed built from the collection of datasets assembled by Zhang et al., but re-doing dataset selection and integration from scratch. The CD8 T cell reference was built from the dataset collection found at utility, as also briefly described in the Readme of ProjecTIL's GitHub. If you are interested, the code for the integration and annotation of the CD8 T cell reference can be found at the following repository: https://github.com/carmonalab/CD8_human_TIL_atlas_construction

Best -massimo