search

epaaso / sc-luca-explore

Exploring the luca dataset for building coabundance networks
0 stars 0 forks source link

How to integrate a new dataset? #22

Open epaaso opened 2 weeks ago

epaaso commented 2 weeks ago

This seems like the best candidate: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11116453/

We will have to use scVI and we have no gpu so it will take a while.

epaaso commented 1 week ago

More candidates and we already implemented Deng: Possible new datasets they are the lung datasets columns

68 zuani & cvejic 9 I-II 77 song & zhang 4 I-II 81 trinks & bishoff 2 or 4 I-II (4 si tomamos los conservados) 85 yang & zhou I-II 7 94 hanley & thomas I-II 9 103 X Deng & Liu 2024 I-II 43 143 Xing & Wang 2021 I-IV 25 168

epaaso commented 1 week ago

The file nb_annot/deng.ipynb has steps to do it with the scLUCA reference atlas. We checked for correctness with normal adjacent samples and it predicts a negligible amount of tumor cells. Though we have only 10 reduced dims while HCLA has 30.