Dealing with duplicated features

[wudustan]

In Read10X(), Seurat allows you to make feature names unique by adding .1, .2, .3 etc at the end of duplicated features. However, subsequenctly they are treated as different genes.

Does it make sense to collapse them into a mean, or take the highest value for each? Some of the features in my data correspond to protein coding genes so this seems like not a trivial issue.

[xiaowenchenjax]

I also met this issue. How to deal with duplicate genes? shall we make unique gene set before creating seurat object? any suggestions?
thanks.

[qslin]

I have this issue, too. How to find the corresponding ensembl ID?

[batalha23]

Hi, I may be a little late to your question but if you look into your features.tsv file (CellRanger output) the first column is EnsemblID and the second is gene name. By default, Seurat's function Read10x() uses the second column, but you can call it with gene.column = 1 to have your transcripts annotated with EnsemblID instead.

[Hu-Clala]

Same issue :(
Is this an issue with the initial GTF file annotation? Does anyone know how to solve it?