search

gbritt / SatijaLab

Single Cell Analysis Scripts
1 stars 0 forks source link

effect of dataset integration on violin plots in Seurat 3 #1

Open justin-malin opened 5 years ago

justin-malin commented 5 years ago

Hello,

Starting with 10X data from three experiments, we followed the integration vignette to align three datasets (code attached). For some cells, this resulted in displayed gene expression levels less than zero (sample plot attached). Is this expected behavior? The violin plots in the Seurat vignettes never show expression < 0. Oddly, gene expression levels for cells in the third dataset (circled in red in attached plot) never display as < 0 for any gene. Normalization and other pre-processing steps are identical for all three datasets.

Is this plausible? Or are we, unwittingly, plotting a mixture of data types (raw, normalized, and/or scaled)?

Thank you!

Justin

Screen Shot 2019-07-12 at 4 17 36 PM

code.txt

justin-malin commented 5 years ago

Should we infer that if no one else has apparently experienced this issue that the source of the problem is our 10X data?

We appreciate your help with this!