Trouble with Ensembl statistics

Question

[mailard]

Hello!

I need to know how many coding genes are in the Y chromosome. The first i try to filter gtf file with R using this code

#load gtf file
gtf <- rtracklayer::import('~/lapd/Index_hum/ann/Homo_sapiens.GRCh38.104.gtf')
gtf_df=as.data.frame(gtf)
##filter gtf file
library(dplyr)
gtf_filt= filter(gtf_df, type=='gene', gene_biotype == 'protein_coding')
chrY=filter(gtf_filt, chromosome_name == 'Y')

Thus, i found 46 coding genes. I thought that i had mistakes and try bioMart:

library(biomaRt)
ensembl = useMart("ensembl", dataset= "hsapiens_gene_ensembl")
new=getBM(attributes=c("chromosome_name","ensembl_gene_id"), filters='biotype', values=c('protein_coding'), mart=ensembl)
 chrY=filter(new, chromosome_name == 'Y')

And found 46 coding genes too. When I try to compare a number of protein-coding genes in ensemble statistics and my annotation file from ensemble i found the second trouble. In Ensembl statistics (http://www.ensembl.org/Homo_sapiens/Location/Genome?r=MT) 20,442 coding genes and in annotation files 19,937.

When do I have mistakes? Or its normal)

[GenoMax]

Y chromosome shows `64 protein coding genes.

Answer 1

[Emily]

Do you mean this page which is the summary for Y, stating 64 coding genes on Y.

The discrepancy of 18 genes is due to the PAR. The genes on the PAR are stored on the X chromosome then the underlying code makes them appear on Y on the website and will come through when you use the APIs. However when you make direct searches of Y through BioMart or in the GTF files, you will only get the non-PAR genes.

[mailard]

Thank you!