14 Getting cell IDs from reference

Natian

With Natian (through Seurat), it is possible to get the cell labels from another Seurat object. This allows users to create a reference dataset or use a publicly available reference dataset and use the identified cell labels to classify cells in a new seurat object.

Key things to remember when using a reference data set

  • Ensure that the reference Seurat object is processed using the same version of Seurat as Natian (version 4.0)
  • Experimental details of the reference data should match as close as possible to the query (or new) Seurat object: For example: Attempting to label SmartSeq data with 10X droplet based methods would result in poor results, because SmartSeq technology usually detects more genes than dropseq methods.
  • For high confidence annotation, the reference should have approximately the same number of more cells than your query.
  • Ensure the gene annotation version used for generating the alignment (processing raw data using Cell Ranger or other pipeline) is the same. Substantial differences in the gene names used in reference compared to the query, could result in poor identification of similar cells.
  • Ensure at least PCA analysis has been performed on the reference data. This can performed by resuming processing of the reference data.
  • Ensure that the cell ids that you wish to transfer are set as Identities of the reference Seurat object

To use reference data set to identify cell ids, Get IDs from reference Seurat button on the left panel. This will open a drop down menu to select a reference Seurat object. Select the reference object and click transfer IDs.

If you do not have a reference, click on the Start Panel , then press the Resume processing Seurat file . This will allow you to browse and load the file.

Once, the transfer is complete the dimensionality plot will be updated to show the new labels.