Genome reannotation of the lizard Anolis carolinensis based on 14 adult and embryonic deep transcriptomes

Abstract

Background: The green anole lizard, Anolis carolinensis, is a key species for both laboratory and field-based studies of evolutionary genetics, development, neurobiology, physiology, behavior, and ecology. As the first non-avian reptilian genome sequenced, A. carolinesis is also a prime reptilian model for comparison with other vertebrate genomes. The public databases of Ensembl and NCBI have provided a first generation gene annotation of the anole genome that relies primarily on sequence conservation with related species. A second generation annotation based on tissue-specific transcriptomes would provide a valuable resource for molecular studies.

Results: Here we provide an annotation of the A. carolinensis genome based on de novo assembly of deep transcriptomes of 14 adult and embryonic tissues. This revised annotation describes 59,373 transcripts, compared to 16,533 and 18,939 currently for Ensembl and NCBI, and 22,962 predicted protein-coding genes. A key improvement in this revised annotation is coverage of untranslated region (UTR) sequences, with 79% and 59% of transcripts containing 5' and 3' UTRs, respectively. Gaps in genome sequence from the current A. carolinensis build (Anocar2.0) are highlighted by our identification of 16,542 unmapped transcripts, representing 6,695 orthologues, with less than 70% genomic coverage.

Conclusions: Incorporation of tissue-specific transcriptome sequence into the A. carolinensis genome annotation has markedly improved its utility for comparative and functional studies. Increased UTR coverage allows for more accurate predicted protein sequence and regulatory analysis. This revised annotation also provides an atlas of gene expression specific to adult and embryonic tissues.

Figure 1

A. Diagram of the bioinformatic pipeline for the A. carolinensis reannotation. B. Venn diagram illustrating the sources of data for the A. carolinensis reannotation. Ab initio, algorithm based gene predictions using Augustus and SNAP [26-28]. RefSeq, alignments of zebrafish, Xenopus frog, chicken, mouse, and human protein and available vertebrate transcripts to the Anocar2.0 genome assembly. NCBI/Ensembl, combined data of A. carolinensis genome annotations from NCBI ref_Anocar2.0 and Ensembl Build 65. RNA-Seq, transcriptomic data from analysis of 14 adult and embryonic tissues.

Figure 2

Increased N50 transcript length and number of predicted transcripts in the ASU annotation.A. The distribution of transcript lengths is shown for the ASU, NCBI and Ensembl genome annotations. The ASU annotation transcript N50 length of 5,355 bp is greater than values for the first generation annotations from Ensembl (2,037 bp) and NCBI (2,364 bp). B. A boxal plot showing the median (horizontal line) and boundaries for the 25^th and 75^th percentiles (box) as well as the range for the ASU, NCBI, and Ensembl predicted transcripts. C. The Notch ligand dll1 is an example of gene whose annotation has been markedly improved in the ASU annotation.