theislab/scib-reproducibility

Reproducibility of simulation data

Closed this issue · 3 comments

demar01 commented

Hello,
First of all, thank you for this fantastic package!

When I generate the simulated data sim1 from here that Anndata does not contain a 'counts' layer in 'adata.layers[counts]', and therefore I cannot use it for the integration part

Would it be possible to get the steps to generate the dataset used in the integration part,
'/storage/groups/ml01/workspace/scIB/simulations_1_1/sim1_1_norm.h5ad'? Also that data should not be normalized, but the name is sim1_1_norm.h5ad?

Many thanks!

LuckyMD commented

Hey! We ran scran normalization on the data to normalize it and put the data that was initially in adata.X in adata.layers['counts']. The function for that should just be scib.pp.normalize() from the scib package here. That should generate exactly the version of the simulation data you need. Maybe @lazappi can confirm this.

LuckyMD commented

Hey @demar01
@lazappi just highlighted the simulation datasets were not on figshare, so I added them. You can find them here.

demar01 commented

Thanks so much!!