Dear all,

'd appreciate also having your suggestions on the following case of scRNAseq analysis with Seurat 3.1. ; the question is : what analysis strategy would you recommend (described below) ?

shall we have 4 batches of scRNA-seq data of these experiments :

WT_batch1, WT_batch2, A_batch1, A_batch2

WT_batch3, WT_batch4, B_batch3, B_batch4

what is the optimal way to analyze the data-sets ? Several analysis strategies are possible :

STRATEGY A.

1. to use CELLRANGER AGGR (with NORMALIZATION = TRUE) on :

WT_batch1, WT_batch2 : to produce WT_batch_1_2

A_batch1, A_batch2 : to produce A_batch_1_2

WT_batch3, WT_batch4 : to produce WT_batch_3_4

B_batch3, B_batch4 : to produce B_batch_3_4**

2. and to follow the descriptions of SEURAT pipelines with LIGER, HARMONY, CONOS, fastMNN (below) on WT_batch_1_2, WT_batch_3_4, A_batch_1_2, B_batch_3_4 :

https://htmlpreview.github.io/?https://github.com/satijalab/seurat.wrappers/blob/master/docs/conos.html

https://htmlpreview.github.io/?https://github.com/satijalab/seurat.wrappers/blob/master/docs/harmony.html

https://htmlpreview.github.io/?https://github.com/satijalab/seurat.wrappers/blob/master/docs/liger.html

https://htmlpreview.github.io/?https://github.com/satijalab/seurat.wrappers/blob/master/docs/fast_mnn.html

STRATEGY B.

1. to use SEURAT MERGE function in order to have all the data (WT_batch1, WT_batch2, A_batch1, A_batch2, WT_batch3, WT_batch4, B_batch3, B_batch4) in a large MATRIX

2. follow the tutorials on SCTRANSFORM, or reciprocal_PCA in https://satijalab.org/seurat/v3.1/integration.html

however, when I call the function FindMarkers in SEURAT (FindMarkers(object, ident.1, ident.2), how shall I specify the REPLICATES in ident.1 and in ident.2 ?

Any other analysis strategy ? Any suggestions, comments would be very welcome ! Thanks a lot !

-- bogdan