How to setup with multiple samples per condition

[ChangqingW]

We have multiple treatment group and multiple samples per treatment group. I was wondering how we should organise the seuList for CreatePRECASTObject. Do we merge the samples by treatment group and have one Seurat object per treatment group? Will the batch effect of spots from the same sample be taken into account if we merge samples?

[feiyoung]

There is no requirement to merge the samples into one Seurat object based on the treatment groups. When utilizing the IntegrateSpaData() function, it is crucial to specify the argument covariates_use and set it to the treatment status. This will ensure proper integration while considering the treatment information for the analysis.

[mei-du]

Hi @feiyoung, I've added the treatment status to each Seurat object's metadata - 4 samples in 1 group, 2 samples in another group. I then followed the tutorial and created the PRECAST object and fitted the data, then integrated all samples. I used "Treatment" with covariates_use but the downstream results appear to be the same as without including covariates. Please advise, thanks!

[feiyoung]

You can verify whether seuInt[['PRE_CAST']]@data is the same for both scenarios.

[Ryeeeeeeeeee]

Hello @feiyoung , I also meet the same problem as @mei-du . However I achieved group comparisons by matching seuInt with the input sample barcode. But I wonder if adding groups via the covariates_use function will have a large impact on the results.

[feiyoung]

In our paper, the analyzed datasets were not from the case-control design, i.e., no treatment group and control group, so we do not explore this impact. However, we do advocate for the incorporation of group information during the integration process, achieved by including the 'covariates_use' parameter. Additionally, you can also try two methods, then see the difference.