Skip to contents

Extract genomic features from a GTFParquet object

Source: R/gtf_parquet.R
genes.Rd

Methods to extract genomic features from a GTFParquet object as GRanges. Unlike TxDb methods, these preserve all GTF attributes as metadata columns.

Usage

# S4 method for class 'GTFParquet'
genes(x, columns=NULL, filter=NULL, use_versioned_ids=FALSE)

# S4 method for class 'GTFParquet'
transcripts(x, columns=NULL, filter=NULL, use_versioned_ids=FALSE)

# S4 method for class 'GTFParquet'
exons(x, columns=NULL, filter=NULL, use_versioned_ids=FALSE)

# S4 method for class 'GTFParquet'
cds(x, columns=NULL, filter=NULL)

# S4 method for class 'GTFParquet'
transcripts(x, columns = NULL, filter = NULL, use_versioned_ids = FALSE)

# S4 method for class 'GTFParquet'
exons(x, columns = NULL, filter = NULL, use_versioned_ids = FALSE)

# S4 method for class 'GTFParquet'
cds(x, columns = NULL, filter = NULL)

Arguments

x

A GTFParquet object.

columns

Character vector of columns to include in mcols. If NULL (default), includes all available attribute columns. For genes: gene_name, gene_type, source, level, tags, havana_gene. For transcripts: transcript_name, transcript_type, gene_id, gene_name, transcript_support_level, ccdsid, protein_id.

filter

Optional named list for filtering features. Names should be column names, values are vectors of acceptable values. Example: filter = list(gene_type = "protein_coding", chrom = "chr1")

use_versioned_ids

Logical. If TRUE, use full versioned IDs (e.g., ENSG00000141510.18). If FALSE (default), use stripped IDs (e.g., ENSG00000141510).

Value

A GRanges object with:

  • Feature IDs as names

  • Genomic coordinates (seqnames, ranges, strand)

  • Genome build in seqinfo (e.g., "GRCh38")

  • Rich metadata in mcols

Details

These methods return GRanges objects with feature IDs as names and rich metadata columns from the original GTF file.

The filter argument enables efficient server-side filtering through Arrow/Parquet predicate pushdown, which can dramatically improve performance compared to subsetting after loading.

Available filter columns include:

  • chrom: Chromosome name

  • gene_type: Gene biotype (e.g., "protein_coding", "lncRNA")

  • transcript_type: Transcript biotype

  • level: Annotation confidence (1=verified, 2=manual, 3=automatic)

  • source: Annotation source ("HAVANA", "ENSEMBL")

See also

Examples

if (FALSE) { # \dontrun{
gtf <- GTFParquet(system.file("gc49", package="lkparq"))

# Extract all genes with full attributes
gr <- genes(gtf)
mcols(gr)  # gene_name, gene_type, level, tags, source, havana_gene

# Filter by gene type
pc <- genes(gtf, filter = list(gene_type = "protein_coding"))
lnc <- genes(gtf, filter = list(gene_type = "lncRNA"))

# Combine filters
pc_chr1 <- genes(gtf, filter = list(gene_type = "protein_coding", chrom = "chr1"))

# Select specific columns only
gr <- genes(gtf, columns = c("gene_name", "gene_type"))

# Use versioned IDs
gr <- genes(gtf, use_versioned_ids = TRUE)
names(gr)[1]  # "ENSG00000141510.18"

# Transcripts with support level
tx <- transcripts(gtf)
high_conf <- tx[mcols(tx)$transcript_support_level == "1"]

# Exons
ex <- exons(gtf, filter = list(chrom = "chr1"))

# CDS with protein IDs
cds_gr <- cds(gtf)
mcols(cds_gr)$protein_id
} # }