gene names not correspond to human ensembl v91 about grein HOT 9 CLOSED

Hello,

This is because some of the datasets were processed using Ensemble v87. We updated the version later on.

Thanks,
Naim

from grein.

Hi team,

Thanks for your reply. Unfortunately, I also checked ensembl v87 and didn't find these genes named "COX1","ND5","COX3","CYTB","COX2" and more, which show up in GSE58375_GeneLevel_Raw_data.csv. I was thinking that if Ensembl v91 gtf is applied to processed data recently so I checked newly processed data (a month ago) such as GSE97615 and GSE115348 and the situation is the same. Could you please check it again? Thank you.

Best,
Zheng

from grein.

Hi Zheng,

This is actually v86 (sorry for the typo!). Our back-end Bioconductor version of Ensembl DB was 86. That's why the discrepancy. We are in the process of upgrading to v91.

Best,
Naim

from grein.

Hi team,

Thanks for your reply. unfortunately I checked ensembl v86 and the situation remains. Could you please check your gtf version again? Thanks.

Best,
Zheng

from grein.

Hi Zheng,

So these are alternative gene symbols available in the AnnotationDbi package. Please check with the corresponding Ensemble id of the genes. I just checked and they are all there. Please let me know if you cannot find them.

from grein.

Hi team,

Thanks for your reply. I actually downloaded ensembl v86(along with v91 and v87) human gif from ensembl and found these genes don't belong to any part of ensembl v86/v91/v87 gtf file. And why did you use AnnotationDbi packages? To convert ENSG to gene symbol? If so, why don't you just extract ENSG-gene_symbol pairs and convert ENSG to gene symbol based on corresponding gtf file? Even though, I still can't find any trace in ensembl v91/v87/v86 gtf by searching the corresponding ENSG ID. It's disturbing that we cannot know which gtf file you actually use, whether the gene names are correct(It has huge effects on downstream analysis), and thus, the credibility of this pipeline should be concerned.

Best,
Zheng

from grein.

As I said, we are using Ensembl version 86. I used "Homo_sapiens.GRCh38.86.gtf" file and double-checked with the ensemble ids. For example, search with ENSG00000198804 for COX1 (alternative symbol MT-CO1). Even if you simply google with the ensembl id, you will see the symbols.
Here, a snapshot of v86 for ensg id ENSG00000198804:

seqnames start end width strand source type score phase gene_id gene_version gene_name gene_source gene_biotype
2574686 MT 5904 7445 1542 + insdc gene NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding
2574687 MT 5904 7445 1542 + ensembl transcript NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding
2574688 MT 5904 7445 1542 + ensembl exon NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding

Since, it is an automated pipeline, AnnotationDbi is the easiest/quickest way for annotation conversion rather than reading the huge gtf file every time.

from grein.

from grein.

As I said, we are using Ensembl version 86. I used "Homo_sapiens.GRCh38.86.gtf" file and double-checked with the ensemble ids. For example, search with ENSG00000198804 for COX1 (alternative symbol MT-CO1). Even if you simply google with the ensembl id, you will see the symbols.
Here, a snapshot of v86 for ensg id ENSG00000198804:

seqnames start end width strand source type score phase gene_id gene_version gene_name gene_source gene_biotype
2574686 MT 5904 7445 1542 + insdc gene NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding
2574687 MT 5904 7445 1542 + ensembl transcript NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding
2574688 MT 5904 7445 1542 + ensembl exon NA NA ENSG00000198804 2 MT-CO1 insdc protein_coding

Since, it is an automated pipeline, AnnotationDbi is the easiest/quickest way for annotation conversion rather than reading the huge gtf file every time.

Thank you for your reply. your answer really helps me. Last question. When do you consider mapping old version alignment RNA-Seq to an uniform genome version? And can we know which dataset is mapped to ensembl v86 or ensembl v91?

Zheng

from grein.