Chromosome-scale assembly of the Verbenaceae species Queen’s Wreath (Petrea volubilis L.)

Abstract

Objectives

Petrea volubilis, a member of the Order Lamiales and the Verbenaceae family, is an important horticultural species that has been used in traditional folk medicine. To provide a genome sequence for comparative studies within the Order Lamiales that includes important families such as Lamiaceae (mints), we generated a long-read, chromosome-scale genome assembly of this species.

Data description

Using a total of 45.5 Gb of Pacific Biosciences long read sequence, we generated a 480.2 Mb assembly of P. volubilis, of which, 93% is chromosome anchored. Representation of genic regions was robust with 96.6% of the Benchmarking of Universal Single Copy Orthologs present in the genome assembly. A total of 57.8% of the genome was annotated as a repetitive sequence. Using a gene annotation pipeline that included refinement of gene models using transcript evidence, 30,982 high confidence genes were annotated. Access to the P. volubilis genome will facilitate evolutionary studies in the Lamiales, a key order of Asterids that includes significant crop and medicinal plant species.

Objective

The Asterid species, Petrea volubilis L., also known as Queen’s Wreath, Purple Wreath, Bluebird vine or Sandpiper vine, is a member of the Verbenaceae family within the Order Lamiales. As a perennial woody vine, P. volubilis is a key ornamental species due to its intense violet flowers. Historically, leaves of P. volubilis have been used in Mexico as folk medicine to remedy kidney stones, rheumatism, diarrhea, and urinary infections [1] and as an abortifacient in Jamaica [2]. P. volubilis extracts have been found to have antipyretic, analgesic, and anti-microbial [3, 4] and insecticidal activities [4]. Recently, P. volubilis was included as one of four outgroup species in a study that revealed the evolutionary basis of chemical diversity in the Lamiacaeae [5]. In this project, we sequenced and annotated the P. volubilis genome to facilitate our understanding of genome and chemodiversity evolution within the Lamiales.

Data description

High molecular weight DNA was isolated using a modified cetyl trimethylammonium bromide method (2% CTAB, 100 mM Tris, 1.4 M Sodium Chloride, 20 mM EDTA) [6] followed by RNase treatment and cleanup using the DNeasy PowerClean Pro Cleanup Kit (Qiagen). Pacific Biosciences (PacBio) SMRTbell Express Template libraries were constructed and sequenced on a PacBio Sequel instrument generating 45.5 Gb of total sequence (Table 1, Data file 1, Data sets 1 & 2, [7]). Reads less than 5 kb were filtered out and the remaining reads were assembled using Canu v1.8 [8] with the options: minOverlapLength = 2000 minReadLength = 5000 genomeSize = 450 m resulting in an initial assembly of 630.0 Mb with 6,515 contigs and an N50 contig length of 369,179 bp. The genome was polished with two rounds of GCpp (v1.9.0) [9], followed by three rounds of polishing with Pilon (v1.23) [10] using Illumina whole genome shotgun reads (Table 1, Data file 1, Data set 3, [7, 11]). A k-mer distribution plot using GenomeScope [12] revealed the genome was heterozygous (Table 1, Data file 2, Data set 3, [7]). Haplotigs were removed using two rounds of purge_dups using the default parameters (v1.0.0) [13, 14] and Hi-C libraries constructed by Phase Genomics (Table 1, Data file 1, Data sets 4 & 5, [7, 15, 16]) were used to place the final scaffolds into 17 chromosomes using the Juicer (v1.6)/3D-DNA pipeline (git commit: 529ccf4; Table 1, Data file 3) [7, 17, 18]. The final assembly size is 480.2 Mb (478.8 Mb ungapped, 93% chromosome-anchored), consistent with the size estimated by flow cytometry of 455 Mb per 1C [5] (Table 1, Data files 4 & 5, [7]). A comparison of k-mers in the Illumina whole genome shotgun reads vs the genome assembly using KAT (v2.4.1) [19] with a k-mer size of 21 revealed that P. volubilis is heterozygous (estimated heterozygosity rate 1.45%) and the assembly is near-complete (estimated completeness, 98.8%;(Table 1, Data file 6, [7]). The majority of k-mers in the reads are present in one copy indicating the haplotigs were successfully purged from the final assembly (Table 1, Data files 1 & 6, Data set 3, [7]). Assessment of representation of genic regions using the Benchmarking of Universal Single Copy Orthologs [20] (BUSCO; v5.4.3 with embryophyta_odb10) revealed 96.6% of the BUSCO genes present in the genome assembly (Table 1, Data file 7, [7]). While the scaffold N50 was 25.6 Mb, the contig N50 was 0.53 Mb due potentially to heterozygosity that reduced the ability of the assembler to generate longer contigs (Table 1, Data file 6, [7]; see Limitations).

Table 1 Overview of data files and data sets used in this study

The P. volubilis genome was annotated as described previously [29]. In brief, repetitive sequences were identified in the unscaffolded contigs using RepeatModeler (v2.0.1) [30] and protein-coding genes removed from the library using ProtExcluder (v1.2) [31]. The custom repetitive sequences were then added to the Repbase Viridiplantae repeats (v20150807) [32] and used to mask repeats using RepeatMasker (v4.1.0) [30] with the parameters -s -nolow -no_is -gff (Table 1, Data file 8, [7]); 57.8% of the genome was masked. RNA-seq reads from five libraries (Table 1, Data file 1, Data sets 6, 7, 8, 9, & 10, [7, 23,24,25,26,27]) were cleaned with Cutadapt (v2.9) [33] using a quality cutoff of 10 and a minimum length 100 nt and then aligned using HISAT2 (v2.2.0) [34] with a maximum intron length of 5000 bp. Gene predictions were generated with BRAKER2 (v2.1.5) [35] using the RNA-seq alignments as hints. Final gene models were refined using the RNA-seq transcript assemblies using two rounds of PASA2 (v2.4.1) [36, 37] and genome-guided transcript assemblies created from the RNA-seq alignments using Stringtie (v2.1.1) [38]. Gene models were annotated using alignments to the predicted Arabidopsis thaliana proteome, Pfam database, and transcript evidence as described previously [29]; a total of 49,169 high confidence models (30,982 genes) within the 56,052 working models (37,610 genes) were annotated (Table 1, Data file 9, [7]). High confidence models within the working model set were defined by either protein evidence (alignment to Arabidopsis or Pfam domain and/or expression evidence (TPM > 0). Representative models, both working and high confidence, were defined as the model for each locus (gene) with the longest CDS. BUSCO assessments (v5.4.3 and embryophyta_odb10) of the annotation revealed 89.9% and 88.5% of BUSCO genes in the working gene model and representative high confidence gene model set, respectively (Table 1, Data file 7, [7]). The final genome annotation was transferred from the scaffolds to the chromosomes using Liftoff (v1.6.3) [39] with the parameters -a 0.9 -s 0.95 -exclude_partial -cds -polish.

Limitations

Petrea volubilis is heterozygous and we purged haplotigs in the assembly process. This likely contributed to the reduced N50 contig size (0.53 Mb) and the slightly larger assembly size (480.2 Mb) compared to the estimated genome size from flow cytometry (445 Mb). However, based on BUSCO scores, a mere 4.3% of the orthologs were duplicated in the assembly suggestive that we removed the majority of alternative haplotigs. Future efforts using near-perfect long genomic reads such as PacBio HiFi or Oxford Nanopore Technologies Q20 + platforms would permit a haplotype-resolved genome assembly.