Is it possible that MNN method can remove "batch" effect between 3' UMI data and full-length read-based data ?

MarioniLab / MNN2017

Code for the MNN manuscript figures

50 stars 19 forks source link

Is it possible that MNN method can remove "batch" effect between 3' UMI data and full-length read-based data ? #8

Closed ahy1221 closed 6 years ago

ahy1221 commented 6 years ago

Such a feature would be quite useful for combing 10X/dropseq data with smartseq2 data. Since the MNN method implicitly performs cell alignment using MNN pairs between different batches, I think the challenge would be how to normalize the data for MNN pairs searching.

LTLA commented 6 years ago

Yes, I think we did this in the paper, with the 68K PBMCs and another T cell dataset.