BRAKER

Overview

BRAKER is a gene prediction pipeline for eukaryotic genomes that combines GeneMark-ES/ET/EP with AUGUSTUS to produce accurate gene structure annotations. It can use RNA-seq alignments, protein homology evidence, or both to train and refine ab initio gene models. BRAKER automates the training of species-specific parameters for AUGUSTUS, making it accessible for newly sequenced organisms without existing gene models. It is one of the most widely used tools for eukaryotic genome annotation and handles intron-exon boundary prediction, alternative splicing hints, and UTR annotation.

Installation

mamba install -c bioconda braker3

Basic Usage

Run BRAKER with RNA-seq evidence provided as a BAM alignment file.

# BRAKER with RNA-seq evidence
braker.pl --genome=genome.fasta \
  --bam=rnaseq_aligned.bam \
  --species=myspecies \
  --softmasking \
  --cores 8

Key Parameters

Flag / option

Description

--genome

Input genome assembly in FASTA format.

--bam

RNA-seq reads aligned to the genome in BAM format (for evidence-based training).

--prot_seq

Protein sequences in FASTA format for homology-based gene finding.

--species

Species name used for AUGUSTUS parameter training and storage.

--softmasking

Indicate that the genome has been soft-masked (lowercase letters for repeats). BRAKER will respect the masking during gene prediction.

--cores

Number of CPU threads to use.

--gff3

Output gene predictions in GFF3 format (default is GTF).

--UTR

Enable UTR prediction (requires RNA-seq evidence).

--fungus

Use fungal-specific parameters for GeneMark and AUGUSTUS.

--AUGUSTUS_ab_initio

Also produce ab initio predictions (without evidence support) in addition to the evidence-based predictions.

Expected Output

BRAKER writes output to the braker/ directory (or the directory specified with --workingdir):

  • braker.gtf – the final gene predictions in GTF format, containing gene, transcript, exon, CDS, and optionally UTR features.

  • braker.gff3 – gene predictions in GFF3 format (when --gff3 is used).

  • braker.aa – predicted protein sequences in FASTA format.

  • braker.codingseq – coding nucleotide sequences in FASTA format.

  • hintsfile.gff – the compiled hints file used by AUGUSTUS, derived from RNA-seq and/or protein evidence.

  • augustus.hints.gtf – AUGUSTUS predictions using the trained parameters and external hints.

See Also

  • Prokka – annotation tool for prokaryotic genomes (bacteria and archaea)

  • Bakta – alternative prokaryotic annotator with comprehensive database support

  • BUSCO – evaluate annotation completeness using conserved gene sets