search

dviraran / SingleR

SingleR: Single-cell RNA-seq cell types Recognition (legacy version)
GNU General Public License v3.0
274 stars 98 forks source link

Seurat object as input and normalization #138

Open Diozao666 opened 4 years ago

Diozao666 commented 4 years ago

Hello Dvir, how are you doing?

  1. I would like to use two rds Seurat objects on SingleR, a test and a ref set (in my own computer, not the webapp) instead of using two separate files for counts and identities and creating an object compatible for both (as instructed in the README). Is this possible?

  2. Within the SingleR function, is it possible to use the Seurat format and path identities with object@meta.data@cell.type and expression slots with object@assays?

  3. From what I have read in the original article, it is necessary to normalize counts for UMI, then TPM and then LOG. Is this correct?

Questions 1 and 2 are mostly for convenience and speed.

Best, Diogo

dviraran commented 4 years ago

Hi Diogo,

As noted, this repository is not currently maintained. Please use the bioconductor version -https://bioconductor.org/packages/release/bioc/html/SingleR.html, and https://github.com/LTLA/SingleR for questions and issues. In short, the answer to 1 is yes, of course. 2. I don't undersrand the question, but of course you can use whatever you want. 3. No. If the data is 3' (like 10X) then no need to normalize to TPM. If its full length (Smart-Seq) then you should definitely normalize to gene length.

Best, Dvir