A modified GC-specific MAKER gene annotation method reveals improved and novel gene predictions of high and low GC content in Oryza sativa

Abstract

BackgroundAccurate structural annotation depends on well-trained gene prediction programs. Training data for gene prediction programs are often chosen randomly from a subset of high-quality genes that ideally represent the variation found within a genome. One aspect of gene variation is GC content, which differs across species and is bimodal in grass genomes. When gene prediction programs are trained on a subset of grass genes with random GC content, they are effectively being trained on two classes of genes at once, and this can be expected to result in poor results when genes are predicted in new genome sequences.ResultsWe find that gene prediction programs trained on grass genes with random GC content do not completely predict all grass genes with extreme GC content. We show that gene prediction programs that are trained with grass genes with high or low GC content can make both better and unique gene predictions compared to gene prediction programs that are trained on genes with random GC content. By separately training gene prediction programs with genes from multiple GC ranges and using the programs within the MAKER genome annotation pipeline, we were able to improve the annotation of the Oryza sativa genome compared to using the standard MAKER annotation protocol. Gene structure was improved in over 13% of genes, and 651 novel genes were predicted by the GC-specific MAKER protocol.ConclusionsWe present a new GC-specific MAKER annotation protocol to predict new and improved gene models and assess the biological significance of this method in Oryza sativa. We expect that this protocol will also be beneficial for gene prediction in any organism with bimodal or other unusual gene GC content.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-017-1942-z) contains supplementary material, which is available to authorized users.

Genome-wide association mapping of date palm fruit traits

“…Homology-based evidence, included 7097 ESTs (downloaded from NCBI EST database on February 9, 2017), protein sequences from Uniprot 51 , a date palm proteome [http://qatar-weill.cornell.edu/research/research-highlights/date-palm-research-program/date-palm-genome-data], an oil palm proteome 52 , and the RNA-Seq derived models from above. Ab initio prediction was performed with Augustus (v. 3.0) trained as described in Bowman et al 53 with gene models produced with StringTie 48 (v. 1.3.2), from the RNA-Seq alignments.…”

“…The raw MAKER2 annotation was parsed, removing models containing TE domains and lacking evidence of transcription or the presence of a Pfam domain as described in Bowman et al 53 . With about 1× of non-organellar single-end WGS Illumina reads, a de novo (non assembly-based) repeat library was produced with RepeatExplorer 54 , and parsed as in Copetti et al 55 .…”

The genome of the bowfin (Amia calva) illuminates the developmental evolution of ray-finned fishes

“…The EST2genome function in MAKER v2.31 35 was used identify putative bow n genes based on BLAST 114 and Exonerate 115 transcript alignments. Best scoring genes with an Annotation Edit Distance of 0.2 or less were used to train Hidden Markov Models with SNAP 116 and AUGUSTUS 117 as described 118 . Ensembl and RefSeq protein sequences from other vertebrate species were used as additional evidence: gar (LepOcu1), coelacanth (LatCha1), mouse (GRCm38.p5), chicken (Gallus_gallus-5.0), human (GRCh38.p10), Xenopus (JGI 4.2), anole lizard (AnoCar2.0), zebra sh (GRCz10), medaka (HdrR), arowana (GCA_001624265.1 ASM162426v1), and elephant shark (GCA_000165045.2 Callorhinchus_milii-6.1.3).…”

Updated Genome Assembly and Annotation for Metrosideros polymorpha, an Emerging Model Tree Species of Ecological Divergence

“…Of the 22,960 gene models generated in the initial MAKER run, 4,224 unique gene models with less than 0.2 of annotation edit distance (AED; Eilbeck et al 2009) were used to train the Hidden Markov model (HMM) for SNAP (ver. 2013-02-16; Korf 2004) and AUGUSTUS (Stanke et al 2004) following Campbell et al (2014) and Bowman et al (2017). MAKER was run again to conduct SNAP- and AUGUSTUS-based gene prediction with the trained HMMs.…”