Which slot is Rmagic pulling from a Seurat object?

[dagarfield]

Hello,

Hopefully an easy question, but when running magic (Rmagic) on a Seurat object, which slot is used? Is it counts, data, or scale.data?

Thanks

[scottgigante]

For MAGIC we used the normalized data, so the data slot.

Seurat v2: https://github.com/KrishnaswamyLab/MAGIC/blob/master/Rmagic/R/magic.R#L267

Seurat v3: https://github.com/KrishnaswamyLab/MAGIC/blob/master/Rmagic/R/magic.R#L332

[dagarfield]

Oh, fantastic. Thanks. And presumably for calculating PCAs, you're then pulling from the scale.data slot? (or even from the PCA stored in the Seurat assay?)

[scottgigante]

No worries. Actually re: PCA, we compute this on the normalized data even if it is computed elsewhere -- PCA is not the biggest computational bottleneck and it's easier just to pull one thing from the Seurat object and then run the rest of our algorithm standalone.

[dagarfield]

Ah! Since you’d included the ScaleData step in some of the tutorials, I’d assumed this came into play. But it’s all much clearer now. Thanks!

On Wed 5. Feb 2020 at 22:35, Scott Gigante notifications@github.com wrote:

No worries. Actually re: PCA, we compute this on the normalized data even if it is computed elsewhere -- PCA is not the biggest computational bottleneck and it's easier just to pull one thing from the Seurat object and then run the rest of our algorithm standalone.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/KrishnaswamyLab/MAGIC/issues/182?email_source=notifications&email_token=ACQMKUYKIA4LREZ2WTOX63DRBMWI3A5CNFSM4KQSEHDKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEK5CEVY#issuecomment-582623831, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQMKUZITSFJTHINUWYS74TRBMWI3ANCNFSM4KQSEHDA .

[scottgigante]

This might be my inexperience with Seurat showing through :) From memory, I believe ScaleData was necessary in order to do some of the later steps in the tutorial, but not for MAGIC.

[dagarfield]

Solid. We can consider this closed. Next step....a wrapper for phateR.....

On Wed 5. Feb 2020 at 22:48, Scott Gigante notifications@github.com wrote:

This might be my inexperience with Seurat showing through :) From memory, I believe ScaleData was necessary in order to do some of the later steps in the tutorial, but not for MAGIC.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/KrishnaswamyLab/MAGIC/issues/182?email_source=notifications&email_token=ACQMKU3MOAS6C3OZNSZJLALRBMX3XA5CNFSM4KQSEHDKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEK5DOAI#issuecomment-582629121, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQMKU2IGQDXAMVDCKZGXMTRBMX3XANCNFSM4KQSEHDA .

[scottgigante]

@dagarfield it's not officially supported but if you install from my fork, I have a wrapper. Read more: https://github.com/satijalab/seurat/pull/1172#issuecomment-564782167