A question about Read10X command, help!

[RiolanWang]

Hello everyone,

I need some help with an issue I am facing. I ran "cellranger count" and then loaded my feature-barcode matrix using "readMM". When I ran "CreateSeuratObject", I received a warning message that stated: "Warning: Non-unique features (rownames) present in the input matrix, making unique." Upon checking my "features.tsv.gz" file, I discovered that some genes, such as the mouse Ptp4a1 gene, have two ENSEMBL IDs (i.e. ENSMUSG00000026064 and ENSMUSG00000117310).

However, when I load the feature-barcode matrix using "Read10X", I do not receive any warning messages when running "CreateSeuratObject". I suspect that this is because the default setting for unique.features is TRUE in the Read10X command. I am curious how Read10X decides which feature to keep if a gene has two different ENSEMBL IDs and therefore, its name is not unique.

I would appreciate it if someone could explain the underlying principle behind this. Thank you!