GeneForge v2.0

A Nextflow Pipeline for Eukaryotic Gene Prediction and Functional Annotation GeneForge is a high-throughput Nextflow pipeline designed for the comprehensive structural and functional annotation of eukaryotic genomes. It orchestrates the parallel execution of BRAKER3 and FunAnnotate, evaluates their performance using BUSCO, and provides a unified functional annotation suite.

Features

Dual Annotation Engine: Parallel execution of BRAKER3 (evidence-based) and FunAnnotate (RNA-seq/Protein-guided). Intelligent Selection: Automatically compares predictions via BUSCO scores to select the highest-quality gene set for downstream annotation. Multi-Omics Integration: Supports Short-read RNA-seq (forward, reverse, or unstranded), Long-read RNA-seq (Nanopore/PacBio), and protein homology evidence. Comprehensive Functional Suite: Integrates InterProScan, EggNOG-mapper, Phobius, and SignalP. Containerized & Reproducible: Built using Nextflow DSL2 with full Singularity/Apptainer support for HPC environments.

Prerequisites

• Nextflow: Version ≥ 22.04
• Singularity/Apptainer: Required for containerized execution.
• Hardware: Linux-based system, 16GB+ RAM (32GB+ recommended), 8+ CPU cores.
• Required Proprietary Files:
  • GeneMark: License key (gm_key_64.gz) and Tarball (gmes_linux_64_4.tar.gz).
  • Functional Tools: Phobius (phobius101_linux.tgz) and SignalP (signalp-6.0h.fast.tar.gz).

Installation

# Clone the repository
git clone https://github.com/yourusername/GeneForge.git
cd GeneForge

# Verify Nextflow installation
nextflow -v

Run Modes

GeneForge uses the --mode flag to define the structural annotation strategy. Every mode concludes with an AGAT-based cleanup to resolve overlaps and ensure GFF3 compliance.

Mode	Prediction Strategy	Final Annotation Logic
braker	BRAKER3 only	Uses BRAKER3 as the reference; merges tRNAs from tRNAscan-SE.
funannotate	FunAnnotate only	Uses FunAnnotate as the reference; merges tRNAs from tRNAscan-SE.
both	Dual Engine	Uses the BUSCO winner as the Backbone. Unique, non-overlapping models from the runner-up are added to "complement" the backbone, followed by tRNA integration.

Functional Annotation (Optional)

Triggered by the --func_annotation flag:

• Scope: Includes Phobius, SignalP, EggNOG-mapper, and InterProScan.
• Target: Runs exclusively on the final merged consensus (the output of the modes above).

---

Workflow Overview

The pipeline follows a modular architecture using Nextflow DSL2:

1. tRNA Scanning: TRNASCAN_SE identifies eukaryotic tRNAs.
2. RNA-seq Processing: RNASEQ_PROCESSING aligns FASTQ files and generates BAM/GTF evidence.
3. Gene Prediction:
   • BRAKER_RUN: Evidence-based gene prediction using RNA-seq and protein homology.
   • FUNANNOTATE: Parallel prediction integrating RNA-seq, protein, and tRNA data.
4. Post-processing: BRAKER_POST standardizes formats and prepares outputs for comparison.
5. Comparison: COMPARE_BUSCO evaluates results to select the highest-quality "Backbone."
6. Merge & Complement (MERGE_ANNOTATIONS):
   • The Backbone is complemented with missing models from the alternative tool.
   • tRNA annotations are merged into the consensus.
   • AGAT is used to resolve overlaps and finalize GFF3 coordinates.
7. Functional Annotation: FUNCTIONAL_ANNOTATION (Optional) adds functional descriptors to the finalized consensus.

Usage

GeneForge uses CSV files to manage complex metadata and file paths.

nextflow run GeneForge/main.nf \
  --mandatory_csv mandatory.csv \
  --optional_csv optional.csv \
  --mode both \
  --func_annotation

Input Configuration

1. mandatory.csv Used for core genomic data and species information.

Format: name,species,organism,busco_db,busco_db_fun,genome_masked,genome_unmasked,protein_evidence,genemark_dir

Example:

name,species,organism,busco_db,busco_db_fun,genome_masked,genome_unmasked,protein_evidence,genemark_dir
Cther,Cladocopium thermophilum,other,alveolata_odb10,protists,/path/to/Cther.fasta.masked,/path/to/Cther.fasta,/path/to/Alveolata.fa,/path/to/genemark

2. optional.csv Used for RNA-seq data, databases, and third-party tool directories.

Format: rnaseq_dir,funanno_DB,eggnog_DB,stranded,nanopore_mrna,pacbio_isoseq,gc_probability,func_tool_dir

Example:

rnaseq_dir,funanno_DB,eggnog_DB,stranded,nanopore_mrna,pacbio_isoseq,gc_probability,func_tool_dir
/path/to/RNA_Cther,/path/to/funannotate_DB,/path/to/eggnog_DB,reverse,/path/to/ONT.fastq.gz,/path/to/pacbio.fastq.gz,0.6377,/path/to/tools

Note: For functional annotation, func_tool_dir must contain:

• phobius101_linux.tgz (Phobius tarball).
• signalp-6.0h.fast.tar.gz (SignalP tarball).

Outputs

Results are organized in the results/ directory:

• geneforge/: The final "Best" annotation set (GFF3, Proteins, BUSCO summary).
• braker/ & funannotate/: Raw prediction outputs from each individual tool.
• tRNA_scan/: High-confidence tRNA annotations.
• busco_comparison/: A summary report comparing BRAKER3 vs. FunAnnotate scores.
• functional_annotation/: Integrated results from Phobius, EggNOG, and InterProScan.

License & Attribution

Primary License

This pipeline is licensed under the MIT License.

Third-Party Licenses

GeneForge automates the use of third-party tools. Users are responsible for complying with their respective licenses:

• GeneMark: Proprietary (Academic use only; commercial license required).
• BRAKER3/tRNAscan-SE/BUSCO: GPL-3.0.
• STAR/Samtools: MIT/BSD.
• Individual license files for these dependencies can be found in /third_party_licenses/.

Citations

If you use GeneForge in your research, please cite:

Sharaf, A., & Voolstra, C. R. (2026). GeneForge v2.0: A Nextflow Pipeline for Gene Prediction and Functional Annotation. Zenodo. https://doi.org/10.5281/zenodo.18592773

Acknowledgments

Supported by the Sequencing Analysis (SequAna) Core Facility at the University of Konstanz.

Contact: [abdoallah.sharaf@uni-konstanz.de]