Comparing groups of single cells

[ghost]

Hi Yuanhua,
I want to perform differential splicing of different groups of single cells. Could you provide an example for groups as I did not find this in the manual.
Thanks

[huangyh09]

Hi,
Thanks for using BRIE to detect differential splicing between groups of single cells. Here is a formal suggestion on it:

"For detecting differential splicing between two cell groups, there are two options. The first option is pooling reads from all cells within a group, and quantify splicing as one ensemble sample, which can be simply done by listing all cells with comma separation in brie. Then brie-diff can detect the differential splicing events between any pair of ensemble samples. This option assumes a homogeneous splicing level across the group, and it requires an additional splicing quantification besides the routine cell-level quantifications. Alternatively, one may use BRIE to quantify splicing in individual cells in each group, and then use a standard hypothesis test, e.g. Wilcoxon, to detect significant changes in the point estimates between two cell populations. This procedure requires however a sufficient number of cells in each population to achieve statistical power, and it ignores the uncertainty in the estimate. The original paper used the first option; we are currently developing a hierarchical model to perform differential splicing analyses while considering the uncertainty and heterogeneity within cluster."

In practice, I would recommend the second option if you have suffcient cells after filtering out the cells with 95% confidence interval > 0.3 (for axample) for each splicing event.

Cheers,
Yuanhua

[aopisco]

@huangyh09 do you have a working prototype for the 2nd option? I'd very interested in trying it out

[liuyu988]

@huangyh09 @aopisco Have you got the working prototype ?thanks

[huangyh09]

Thanks. I think the brie-quant in BRIE2 solves it as a generalised linear model, see the manual here.

Yuanhua