I work with biomaRt in R most of the time, but I do use .gtf files for alignment of our NGS data sets. While cross checking some items I noticed that the number of ensembl_gene_id do not match the number of ensembl_gene_id's found in the Homo_sapiens.GRCh38.113.gtf file. Specifically there are 86402 unique ensembl_gene_id's from biomaRt and 78932 unique ensembl_gene_id's in the .gtf file. As an example, Gene NPY4R2 has two different ensembl_gene_id's from biomaRt (ENSG00000288157, ENSG00000264717) and it only has one ensembl_gene_id from the .gtf file (ENSG00000264717).

Furthermore, I have examined rat, mouse, cow and sheep genomes. The rat, mouse, and cow are consistent between the .gtf files and their biomaRt alternative, but sheep actually has more unique ensembl_gene_ids in the .gtf (30722) file than in biomaRt (27054). Furthermore the sheep ensembl_gene_ids follow two different naming schemes. In the Ovis_aries_rambouillet.ARS-UI_Ramb_v2.0.113.gtf file, the COX1 gene comes up as ENSOARG00020000017, where as biomaRt comes up as ENSOARG00000000016.

Any clarification would be greatly appreciated on this matter. Below, I have attached the R code I have used from biomaRt, as well as the python script I used to parse the .gtf files.

#getBM method (Human)
 ensembl <- biomaRt::useMart('ensembl')
  mart <- biomaRt::useDataset("hsapiens_gene_ensembl", ensembl)


  #View(listAttributes(mart))

  BMattribHuman <- biomaRt::getBM(attributes = c('ensembl_gene_id', 'external_gene_name'),#, 'go_id'),
                             filters = "ensembl_gene_id",
                             values = "",
                             mart = mart)

#getBM method (sheep)
 ensembl <- biomaRt::useMart('ensembl')
  mart <- biomaRt::useDataset("oaries_gene_ensembl", ensembl)


  #View(listAttributes(mart))

  BMattribAries <- biomaRt::getBM(attributes = c('ensembl_gene_id', 'external_gene_name'),#, 'go_id'),
                             filters = "ensembl_gene_id",
                             values = "",
                             mart = mart)

sessionInfo()

R version 4.4.0 (2024-04-24)
Platform: aarch64-apple-darwin20
Running under: macOS 15.3.2

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: America/Toronto
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
 [1] rappdirs_0.3.3          generics_0.1.3          xml2_1.3.6             
 [4] RSQLite_2.3.9           stringi_1.8.4           hms_1.1.3              
 [7] digest_0.6.37           magrittr_2.0.3          evaluate_1.0.1         
[10] fastmap_1.2.0           blob_1.2.4              jsonlite_1.8.9         
[13] progress_1.2.3          AnnotationDbi_1.66.0    GenomeInfoDb_1.40.1    
[16] DBI_1.2.3               httr_1.4.7              purrr_1.0.2            
[19] UCSC.utils_1.0.0        Biostrings_2.72.1       httr2_1.0.7            
[22] cli_3.6.3               rlang_1.1.4             crayon_1.5.3           
[25] dbplyr_2.5.0            XVector_0.44.0          Biobase_2.64.0         
[28] bit64_4.5.2             withr_3.0.2             cachem_1.1.0           
[31] yaml_2.3.10             tools_4.4.0             memoise_2.0.1          
[34] dplyr_1.1.4             filelock_1.0.3          GenomeInfoDbData_1.2.12
[37] BiocGenerics_0.50.0     curl_6.1.0              vctrs_0.6.5            
[40] R6_2.5.1                png_0.1-8               stats4_4.4.0           
[43] lifecycle_1.0.4         BiocFileCache_2.12.0    zlibbioc_1.50.0        
[46] KEGGREST_1.44.1         stringr_1.5.1           S4Vectors_0.42.1       
[49] IRanges_2.38.1          bit_4.5.0.1             pkgconfig_2.0.3        
[52] pillar_1.10.1           glue_1.8.0              xfun_0.50              
[55] tibble_3.2.1            tidyselect_1.2.1        rstudioapi_0.17.1      
[58] knitr_1.49              htmltools_0.5.8.1       rmarkdown_2.29         
[61] compiler_4.4.0          prettyunits_1.2.0       biomaRt_2.60.1

Python script below:

# For running on macOS
# python3 -m venv pythonEnv
# source pythonEnv/bin/activate
# pip install pandas



import pandas as pd
import os

# Ask user for file name and location, set to wd
file = input("Enter the file name and path: ")  #"/Users/username/..../Homo_sapiens.GRCh38.113.gtf"
savename = input("Enter the name that you would like to save the file as: ")
filePath = "/".join(file.split("/")[:-1])
print(filePath)
print(f"Does the Path exist? {os.path.exists(file)}")

# Initialize variables
data = []
genebuild_last_updated = None

# Read the file and process lines
with open(file) as f:
    for line in f:
        if line.startswith("#!genebuild-last-updated"):
            genebuild_last_updated = line.strip().split()[-1]
        if line.startswith("#"):  # Skip comments
            continue
        cols = line.strip().split("\t")
        if cols[2] != "gene":
            continue
        attributes = cols[8].split("; ")
        attr_dict = {}
        for attr in attributes:
            if " " in attr:
                key, value = attr.split(" ", 1)
                attr_dict[key] = value.strip('"')
        data.append([attr_dict.get("gene_id"), attr_dict.get("gene_name")])

# Create a DataFrame
df = pd.DataFrame(data, columns=["Ensembl_ID", "GeneName"])
df = df.sort_values(by="Ensembl_ID", ascending=True)
df.drop_duplicates().to_csv(os.path.join(filePath, f"{savename}_LastGeneBuild{genebuild_last_updated}.csv"), sep=",", index=False)

print("Done!")

I think this is because Ensembl can assign different Gene IDs to "the same gene" if it is found both on a standard chromosome and a scaffold that hasn't be placed in a chromosome. In your NPY4R2 example ENSG00000264717 is on chromosome 10 but ENSG00000288157 is on an unplaced scaffold. Presumably they're sufficiently similar / identical for the HGNC to consider them both NPY4R2, but Ensembl choose to assign separate IDs.

I haven't checked, but my guess would be that your GTF files don't contain these unplaced scaffolds, and you'd want Homo_sapiens.GRCh38.113.chr_patch_hapl_scaff.gtf for that.