Improved annotation of the insect vector of citrus greening disease: biocuration by a diverse genomics community

Surya Saha, Prashant S. Hosmani, Krystal Villalobos-Ayala, Sherry Miller, Teresa Shippy, Mirella Flores, Andrew Rosendale, Chris Cordola, Tracey Bell, Hannah Mann, Gabe DeAvila, Daniel DeAvila, Zachary Moore, Kyle Buller, Kathryn Ciolkevich, Samantha Nandyal, Robert Mahoney, Joshua Van Voorhis, Megan Dunlevy, David C. Farrow, David Hunter, Taylar Morgan, Kayla Shore, Victoria Guzman, Allison Izsak, Danielle E. Dixon, Andrew Cridge, Liliana M. Cano, Xiaolong Cao, Haobo Jiang, Nan Leng, Shannon Johnson, Brandi L. Cantarel, Stephen Richards, Adam C. English, Robert G. Shatters, Christopher P. Childers, Mei-Ju Chen, Wayne B. Hunter, Michelle Cilia, Lukas A. Mueller, Monica C. Munoz-Torres, David Nelson, Monica Poelchau, Joshua B. Benoit, Helen Wiersma-Koch, Tom D'Elia, Susan J. BrownDatabase J. Biol. Databases Curation 2017 · DOI

TL;DR

This paper presents a comprehensive, manually curated genome annotation for the Asian citrus psyllid (Diaphorina citri), the vector of citrus greening disease. By combining automated pipelines with a large-scale community-driven biocuration effort involving students and experienced annotators, they've produced a high-quality Official Gene Set (OGS v1.0) that significantly improves our understanding of this economically important pest's biology and immune system.

What Problem Does This Paper Address?

Citrus greening disease, caused by the bacterium Candidatus Liberibacter asiaticus (CLas), is a devastating threat to citrus production worldwide. The primary method of control is suppressing the insect vector, the Asian citrus psyllid (Diaphorina citri). To develop more targeted and sustainable control strategies, a deep understanding of the psyllid's genome, particularly genes involved in its interaction with CLas and its own biological processes, is crucial. While draft genome assemblies and automated annotations exist, they often contain errors and omissions, hindering the identification of effective control targets. This paper addresses the need for a highly accurate and well-annotated genome to facilitate the development of novel molecular-based pest management strategies.

Methods

The authors employed a multi-pronged approach to achieve a high-quality genome annotation:

  • Genome Sequencing and Assembly: High-molecular-weight DNA was sequenced using PacBio and Illumina technologies. The draft assembly (Diaci1.1) was generated using Velvet and scaffolded with PacBio long reads via PBJelly.
  • Transcriptome Assembly (MCOT v1.0): A comprehensive transcriptome assembly was created by merging outputs from multiple gene prediction and transcriptome assembly tools (Maker, Cufflinks, Oases, Trinity). This aimed to capture a more complete set of gene models, especially those potentially missed or misassembled in the draft genome.
  • Automated Annotation: The NCBI Eukaryotic Genome Annotation pipeline (NCBI v100) and the MAKER annotation pipeline (Maker v1.1) were used to generate initial gene models.
  • Community-Based Manual Curation: This was the core innovation. A large, diverse group of undergraduate and graduate students, along with experienced annotators, was trained to manually curate gene models. They used the i5k Apollo platform, integrating various evidence tracks (RNAseq, proteomics, homology) to refine gene structures, assign functions, and ensure consistent naming conventions.
  • Annotation Edit Distance (AED): This metric was used to evaluate the quality of different annotation sets by measuring their congruence with RNAseq evidence.
  • BUSCO Analysis: Benchmarking of Universal Single-Copy Orthologs was performed to assess the completeness of gene sets.

Datasets

  • Diaphorina citri Genome Assembly: Diaci1.1 (Illumina and PacBio data).
  • Diaphorina citri Transcriptome Assembly: MCOT v1.0.
  • NCBI Diaphorina citri Annotation Release 100 (NCBI v100): An existing automated annotation.
  • Maker v1.1 Annotation: An annotation generated using the MAKER pipeline.
  • RNAseq data: From adult, nymph, and egg tissues, as well as CLas-exposed and healthy samples.
  • Homology data: From insect proteins in NCBI nr and Swiss-Prot databases, and specific immune gene databases like ImmunoDB.
  • Orthologous genes: From related insect species (e.g., Drosophila melanogaster, Aedes aegypti, Anopheles gambiae, Cimex lectularius, Acyrthosiphon pisum).

Key Insights and Ideas

  • The Power of Collaborative Biocuration: This paper strongly advocates for community-driven annotation projects, demonstrating that a large, distributed team can effectively produce high-quality annotations, especially when well-trained and guided by standardized protocols. This model democratizes genome annotation and serves as an excellent training platform.
  • MCOT Transcriptome's Value: The MCOT transcriptome assembly proved crucial in identifying genes missing or misassembled in the draft genome, highlighting the importance of complementary transcriptome data for improving genome annotation.
  • Immune System Reduction in Hemipterans: The study found a reduced repertoire of immune genes in D. citri compared to Dipterans, a trend also observed in other hemipterans. This is hypothesized to be linked to their association with microbial symbionts and feeding on relatively sterile plant fluids.
  • RNAi Machinery Completeness: D. citri possesses a full complement of RNAi machinery genes, with duplications observed in genes associated with the miRNA pathway, mirroring findings in pea aphids but differing from whiteflies.
  • P450 Gene Blooms: The identification of "blooms" (expansions) in specific Cytochrome P450 gene families (CYP4, CYP2, and CYP3 clans) suggests an evolutionary response to environmental stresses, potentially related to detoxification or metabolism of plant compounds.

Key Results

  • 530 Manually Curated Gene Models: The community effort successfully curated 530 gene models, focusing on immunity-related genes, RNAi pathways, CYP genes, and other genes relevant to insect development and physiology.
  • Improved Annotation Quality: Manual curation significantly improved gene model quality, as evidenced by higher AED scores for curated genes compared to automated predictions.
  • MCOT v1.0 Completeness: The MCOT v1.0 transcriptome assembly showed higher BUSCO completeness (92.9%) than other available annotation sets, indicating a more comprehensive gene repertoire.
  • Identification of Key Gene Families: The study provided detailed insights into immune pathways (Toll, Imd, JAK/STAT), RNA interference machinery, and Cytochrome P450 genes in D. citri.
  • Official Gene Set (OGS v1.0): The resulting OGS v1.0 integrates both automatically predicted and manually curated gene models, serving as a valuable resource for the research community.

Limitations and Future Work

  • MCOT Mapping to Draft Genome: A significant number of MCOT proteins could not be mapped to the draft Diaci1.1 genome, suggesting potential issues with the draft assembly itself.
  • AED Scores for MCOT: While MCOT v1.0 had higher BUSCO scores, its mapped set scored lower in AED analysis due to mapping challenges.
  • Absence of Certain Genes: The study noted the absence of certain immune genes (e.g., GNBP, some AMPs) which, while surprising, is not unprecedented in hemipterans and may be linked to their unique biology.
  • Future Work: The authors plan to continue this community annotation process with future cohorts of students and scientists to improve the characterization of the D. citri genome for subsequent versions. They also highlight the need for an improved reference genome assembly for D. citri using advanced sequencing technologies. The MCOT transcriptome is available for further research.

Summary updated

BibTeX
@article{saha2017improved,
  author = {Surya Saha and Prashant S. Hosmani and Krystal Villalobos-Ayala and Sherry Miller and Teresa Shippy and Mirella Flores and Andrew Rosendale and Chris Cordola and Tracey Bell and Hannah Mann and Gabe DeAvila and Daniel DeAvila and Zachary Moore and Kyle Buller and Kathryn Ciolkevich and Samantha Nandyal and Robert Mahoney and Joshua Van Voorhis and Megan Dunlevy and David C. Farrow and David Hunter and Taylar Morgan and Kayla Shore and Victoria Guzman and Allison Izsak and Danielle E. Dixon and Andrew Cridge and Liliana M. Cano and Xiaolong Cao and Haobo Jiang and Nan Leng and Shannon Johnson and Brandi L. Cantarel and Stephen Richards and Adam C. English and Robert G. Shatters and Christopher P. Childers and Mei-Ju Chen and Wayne B. Hunter and Michelle Cilia and Lukas A. Mueller and Monica C. Munoz-Torres and David Nelson and Monica Poelchau and Joshua B. Benoit and Helen Wiersma-Koch and Tom D'Elia and Susan J. Brown},
  title = {{Improved annotation of the insect vector of citrus greening disease: biocuration by a diverse genomics community}},
  journal = {Database J. Biol. Databases Curation},
  year = {2017},
  doi = {10.1093/database/bax032},
}
Figure