Redo the BMMC figures

[LTLA]

We should have used a slightly larger radius for the BMMC analysis. For some reason, the default radius gets us about 5 cells a hypersphere, which is not enough for differential testing. It seems that the intensities are quite highly variable across this population of cells - too diverse a mixture, perhaps? - so that even with more cells than the MEF reprogramming data set, the cells per hypersphere are lower.

A value of 0.55 seems appropriate, based on some diagnostic plots. This gets us to a median count of ~15 for each hypersphere, which should be good enough for analysis. Doing so identifies B cells as responding to IL-10 with increased pSTAT3, which is expected.

[jcmarioni]

Ok. This seems reasonable but not critical right? The filtering argument you will have to explain to me. Overall there is nothing too substantial here right? John

…

Sent from my iPhone

On 3 Dec 2016, at 12:22, Aaron Lun ***@***.***> wrote: We should have used a slightly larger radius for the BMMC analysis. For some reason, the default radius gets us about 5 cells a hypersphere, which is not enough for differential testing. It seems that the intensities are quite highly variable across this population of cells - too diverse a mixture, perhaps? - so that even with more cells than the MEF reprogramming data set, the cells per hypersphere are lower. A value of 0.55 seems appropriate, based on some diagnostic plots. This gets us to a median count of ~15 for each hypersphere, which should be good enough for analysis. Doing so identifies B cells as responding to IL-10 with increased pSTAT3, which is expected. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[LTLA]

It's fine. In fact, it doesn't even need to be changed - while the current analysis is suboptimal, it's still pretty good, so I don't think anyone will notice. It's primarily for my own peace of mind.

[LTLA]

Done, updated BMMC analysis is now SF16-17.