search

MarioniLab / FurtherMNN2018

Code for further development of the mutual nearest neighbours batch correction method, as implemented in the batchelor package.
22 stars 6 forks source link

Question about "pancreas/prepareData.R" #8

Closed ssscj closed 5 years ago

ssscj commented 5 years ago

thanks for developing this package While running sce.emtab6<- computeSumFactors(sce.emtab,clusters=clusters1,min.mean=1) it showed

Warning message: In .local(x, ...) : zero spike-in counts during spike-in normalization

and while running: sce.emtab3 <- normalize(sce.emtab2,return_log=FALSE) it showed

Error in .local(object, ...) : size factors should be positive real numbers

how could I fix it? Thank you!

LTLA commented 5 years ago
  1. The warning arises from computeSpikeFactors, not computeSumFactors.
  2. You can skip the normalize() step, everything gets re-normalized in plotCorrections.R anyway.
ssscj commented 5 years ago

Thanks for your reply. I'm sorry for my mistake. I have one more question. Can I skip the computeSpikeFactors() step instead of normalize() step, because I think the spike-ins are not important in the analysis.

LTLA commented 5 years ago

The spike-ins are used for modelling variability in each batch. Now, that's not critical, but we already have the spike-ins in this data set, so why not use them? If you're referring to your own data set that doesn't have spike-ins, you can consider alternative strategies as described here.