MAKER2: an annotation pipeline and genome-database management tool for second-generation genome projects.

Holt C, Yandell M. · 2011

Read MAKER2: an annotation pipeline and genome-database management tool for second-generation genome projects. by Holt C, Yandell M. (2011) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc3280279.txt.

Introduction

Carson Holt and Mark Yandell presented MAKER2, a genome annotation pipeline built for the sequencing era, in this 2011 paper in BMC Bioinformatics. Second-generation projects more and more tackle exotic organisms with no gold-standard gene sets to train prediction software, and cheap RNA sequencing invites re-annotation of published genomes. Existing tools were not designed for either situation.

MAKER2 runs as a parallel, multi-threaded application that scales to datasets of virtually any size. The authors show it produces accurate annotations where training data are limited, poor, or absent, and that folding in RNA-seq evidence improves quality. It can also refresh legacy annotations, manage the underlying genome database, and grade annotation quality, flagging suspect gene models for human review. They present it as the first annotation engine designed for second-generation genome projects.

Catalog record

CREATORHolt C, Yandell M.
YEAR2011
COLLECTIONPAPERS
FORMATTXT
FILE SIZE69 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC3280279

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

MAKER2: an annotation pipeline and genome-database management tool for second-generation genome projects By Carson Holt, Mark Yandell Published in: BMC Bioinformatics DOI: 10.1186/1471-2105-12-491 ABSTRACT Background Second-generation sequencing technologies are precipitating major shifts with regards to what kinds of genomes are being sequenced and how they are annotated. While the first generation of genome projects focused on well-studied model organisms, many of today's projects involve exotic organisms whose genomes are largely terra incognita. This complicates their annotation, because unlike first-generation projects, there are no pre-existing 'gold-standard' gene-models with which to train gene-finders. Improvements in genome assembly and the wide availability of mRNA-seq data are also creating opportunities to update and re-annotate previously published genome annotations. Today's genome projects are thus in need of new genome annotation tools that can meet the challenges and opportunities presented by second-generation sequencing technologies. Results We present MAKER2, a genome annotation and data management tool designed for second-generation genome projects. MAKER2 is a multi-threaded, parallelized application that can process second-generation datasets of virtually any size. We show that MAKER2 can produce accurate annotations for novel genomes where training-data are limited, of low quality or even non-existent. MAKER2 also provides an easy means to use mRNA-seq data to improve annotation quality; and it can use these data to update legacy annotations, significantly improving their quality. We also show that MAKER2 can evaluate the quality of genome annotations, and identify and prioritize problematic annotations for manual review. Conclusions MAKER2 is the first annotation engine specifically designed for second-generation genome projects. MAKER2 scales to datasets of any size, requires little in the way of training data, and can use mRNA-seq data to improve annotation quality. It can also update and manage legacy genome annotation datasets. BACKGROUND Second-generation sequencing technologies are creating new opportunities as well as new challenges for genomics research. While first-generation genome projects focused primarily on established model organisms such as Drosophila melanogaster[1], Caenorhabditis elegans[2], and Mus musculus[3], falling sequencing costs are allowing second-generation genome projects to focus on more exotic and phylogenetically isolated organisms. The large volumes of data produced by second-generation sequencing technologies also create difficulties for data management not encountered by first-generation projects. Together, these factors can spell disaster for second-generation genome projects. The first-generation of genome projects benefited greatly from large bodies of pre-existing knowledge regarding their organisms' genomes. For D. melanogaster, C. elegans, and Homo sapiens[4,5], for example, hundreds of published gene models already existed before these genomes were sequenced. These pre-existing published gene models were critical for annotation and analysis, because they allowed researchers to train and optimize gene prediction and annotation tools for each genome. They also provided a standard by which to judge the accuracy of annotations. Second-generation projects rarely have access to such information. This severely limits their ability to train ab initio gene-finders, the result being (as we show below) low-quality gene predictions. The lack of pre-existing gene models also leaves many second-generation projects with no objective standards with which to gauge annotation accuracy. Quality control is thus a significant issue for these projects; data management is another. New techniques such as high-throughput transcriptome sequencing (mRNA-seq) have great potential to improve annotation quality, but they produce enormous amounts of data; likewise, the existence of legacy annotations for a large number of both first and second-generation genomes is also creating data management challenges. Thus, it is essential that the output of an annotation pipeline be easily converted into a genome database. MAKER2 builds upon MAKER[6], an easy-to-use genome annotation pipeline that has seen wide adoption[7-19]. MAKER2 improves upon the de novo annotation capabilities of the original MAKER and integrates support for multiple ab initio prediction tools. Major additions to MAKER2 include integration of the Annotation Edit Distance (AED)[20] metric for improved quality control and downstream database management, support for mRNA-seq to allow researchers to leverage second generation sequencing technologies, and gene model pass-through capability; thus creating a first of it's kind tool for updating and reannotating existing model organism databases. The pipeline also supports distributed parallelization on computer clusters via MPI which means MAKER2 can scale to datasets of virtually any size. MAKER2 can run on UNIX-like operating systems such as Linux and Darwin in Mac OS X. MAKER2 thus provides straightforward solutions to the problems facing today's second-generation genome projects. Here, we demonstrate MAKER2's ability to overcome handicaps resulting from a lack of pre-existing gene models with which to train gene-predictors for use on novel genomes; its ability to use mRNA-seq data to improve annotation quality; and its ability to update legacy annotations and in doing so significantly improve their quality. MAKER2 is not only an improved annotation engine; it is also a new kind of annotation management tool. Throughout these analyses, we measure MAKER2's performance using the integrated annotation quality-control measure AED, developed by the Sequence Ontology project[21]. We show that MAKER2 can both evaluate the global quality of genome annotations, and identify and prioritize problematic annotations for manual review; these are functionalities offered by no other annotation tool. IMPLEMENTATION De novo annotation of first-generation genomes D. melanogaster chromosome 3R and GFF3 annotations for release r5.32 were downloaded from FlyBase. C. elegans chromosome V and GFF3 annotations for release WS221 were downloaded from WormBase. Arabidopsis thaliana[22] chromosome 4 and GFF3 annotations were downloaded from TAIR. Each set of reference gene annotations were passed to MAKER2's model_gff option with all prediction and evidence alignment options turned off. This has the effect of repackaging the reference gene models into more standardized GFF3 files compatible with downstream analysis scripts. Ab initio gene predictions were produced by the programs SNAP[23] version 2010-07-28, Augustus[24] 2.5.5, and GeneMark-ES[25] 2.3a, using the D. melanogaster, C. elegans, and A. thaliana parameter files pre-packaged with each algorithm (GeneMark parameter files are packaged with the GeneMark.hmm download). To produce all predictions in standardized GFF3 format, these algorithms were run through MAKER2 with all evidence alignments options turned off and the keep_preds flag set to 1. This has the effect of only producing raw ab initio gene predictions in standardized GFF3 format. Evidence-based gene annotations in MAKER2 were produced using default settings. The species parameter files were the same as those used for the ab initio gene-predictors. EST and protein homology datasets were provided for each organism. For D. melanogaster, the EST dataset consisted of all D. melanogaster ESTs available from dbEST[26], and the protein homology input consisted of all Anopheles gambiae[27] proteins from NCBI together with all of the UniProt/Swiss-Prot[28,29] database proteins (minus Drosophila proteins). For C. elegans, the EST dataset consisted of all C. elegans release WS221 ESTs available from WormBase, and protein homology input consisted of all Caenorhabditis briggsae[30] WS221 proteins from WormBase together with all of the UniProt/Swiss-Prot database proteins (minus Caenorhabditis proteins). The EST dataset for A. thaliana consisted of all A. thaliana ESTs from dbEST, and the protein homology dataset consisted of all Oryza sativa[31] release 6.1 proteins from PlantGDB and all of the UniProt/Swiss-Prot database proteins (minus Arabidopsis proteins). For A. thaliana, MAKER2 was also provided with the Arabidopsis transposable element FASTA file available from TAIR (assists in repeat masking). The reference gene models, ab initio gene predictions, and evidence-based gene annotations were converted to GTF format using the maker2eval

Read the complete text