search

drisso / zinbwave

Clone of the Bioconductor repository for the zinbwave package, see https://bioconductor.org/packages/zinbwave
43 stars 10 forks source link

Negative values in normalizedValues output #67

Closed agranado closed 3 years ago

agranado commented 3 years ago

I am trying to use the normalizedValues from Zinbwave for downstream analysis and I noticed that the matrix contains several cell-gene data points with negative values. Is this correct? Is there a way to get normalized counts from zinbwave to correct the dropout and then use them for downstream analysis?

drisso commented 3 years ago

Hi @agranado,

the normalizedValues are returned by zinbwave only for the purpose of data exploration and graphical display, their use in downstream analysis is untested and not recommended. Our suggested workflow is to base downstream unsupervised analyses on the latent factors W and the supervised analyses on edgeR/DESeq2 with weights.

There are many other methods that can be used to correct for dropouts, but not zinbwave.

agranado commented 3 years ago

Hi @drisso Thanks a lot for your reply! I understand so we will keep the W for visualization and clustering.

Thanks a lot.