about the use of CellRanger aggr

Question

[Bogdan]

Dear all,

i am writing to ask please for your suggestions about the appropriate use of CELLRANGER AGGR function in the following context :

we do have 2 experimental replicates for a WT and DISEASE mice :

-- WT_replicate1

-- WT_replicate2

-- DIS_replicate1

-- DIS_replicate2

in order to combine the replicates, shall we just use CELLRANGER AGGR, or shall we just put the fastq files in the same folders i.e.

the fastq files of WT_replicate1, WT_replicate2 in a folder,

and

the fastq files of DIS_replicate1, DIS_replicate2 in another folder ?

thanks a lot,

bogdan

[Bogdan]

that is a great suggestion, thank you very much !

[Bogdan]

Dear Lior, on a side note, may I ask you please for another suggestion :

what batch-correction method would you recommend for scRNA-seq ? CCA, MNNCORRECT, other methods ?

thanks again !

Answer 1

[Lior Pachter]

You should not process the FASTQs together, because cell barcodes will clash. So you should process them separately. I would not recommend merging the replicates after processing, information about variance within condition will be useful and important for downstream analysis. If you do aggregate the datasets, you should be aware that the aggr function performs a normalization of the data. In the kallisto|bustools workflow we have a tutorial for aggregation that allows you to explore other normalizations.

Answer 2

[swbarnes2]

cellranger aggr aggregates outputs from multiple runs of cellranger count, normalizing those runs to the same sequencing depth and then recomputing the feature-barcode matrices and analysis on the combined data. The aggr pipeline can be used to combine data from multiple samples into an experiment-wide feature-barcode matrix and analysis.

https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/what-is-cell-ranger

[Bogdan]

Thank you. I have received also an answer from 10X genomics that say :

1. First you will need to run cellranger count separately four times on each of your 4 libraries:

https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/using/count:

1. Then to combine them with cellranger aggr​ on all four libraries at the same time, I would set up a CSV like this:

library_id,molecule_h5,treatment

WT_replicate1,/path/to/molecule_info.h5,WT

WT_replicate2,/path/to/molecule_info.h5,WT

DIS_replicate1,/path/to/molecule_info.h5,DIS

DIS_replicate2,/path/to/molecule_info.h5,DIS

This way you can combine all four libraries in Loupe browser and combine the WT and DIS treatments if you like.

Please see: https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/using/aggregate