feature.tsv from scRNA-seq dataset has gene name with dash number suffix following adopting NCBI reference

Question

[my.kim]

Hi,

I'm a user of scRNA-seq in veterinary medicine field. like scRNA-seq analysis. Yesterday, I got fatsq files, and using these I ran a cellranger count with the canine reference genome canfam6.0, a newly released one after canfam3.1. For transcriptome reference, I downloaded fasta and gtf files from the NCBI website (https://www.ncbi.nlm.nih.gov/datasets/genome/GCF_000002285.5/), ran the mkref command, and generated a canine reference genome. I used this reference genome to run the cellranger count. After that, I found that feature.tsv contains the suffix "dash number", for example, CD79A-4, PTPRC-5, and even CD3D-13. I was wondering why this phenomenon occurred to me and how to correct this issue.

Any help will be very useful to me.

Thank you.

Kim Sincerely,

[my.kim]

For example, here is the results

Answer 1

[bk11]

These is something off in your input GTF file. You can spot check some gene names how they are coded. Please try using CanFam3.1 (INSDC Assembly GCA_000002285.2, Sep 2011) canine reference genome and cellranger mkgtf filtered CanFam3.1 GTF from Ensembl v102 as input to create canine-specific reference package for cellranger. Please check out this paper-

https://www.biorxiv.org/content/10.1101/2021.06.29.450365v1.full.pdf

[my.kim]

I do appreciate it. I found something you already expected.

When I read NCBI canfam6 gtf file, all genes have the dash number suffix as below. However, canfam3.1 gtf. files from NCBI and Ensembl do have correct gene name. So, can we move the attached suffix in the GTF file?

[bk11]

You could change the attached suffix in the GTF file but I m not sure if the canfam6 gtf is well curated. To be in the safe side, you ll be better off using canfam3.1 gtf file.