Prunus campanulata Genome v1.0 Assembly & Annotation

Publication

Hu, Y., Feng, C., Wu, B., & Kang, M. (2023). A chromosome-scale assembly of the early-flowering Prunus campanulata and comparative genomics of cherries. Scientific Data, 10, 920.

Abstract

Prunus campanulata is an important flowering cherry germplasm of high ornamental value. Given its early-flowering phenotypes, P. campanulata could be used for molecular breeding of ornamental species and fruit crops belonging to the subgenus Cerasus. Here, we report a chromosome-scale assembly of P. campanulata with a genome size of 282.6 Mb and a contig N50 length of 12.04 Mb. The genome contained 24,861 protein-coding genes, of which 24,749 genes (99.5%) were functionally annotated, and 148.20 Mb (52.4%) of the assembled sequences are repetitive sequences. A combination of genomic and population genomic analyses revealed a number of genes under positive selection or accelerated molecular evolution in P. campanulata. Our study provides a reliable genome resource, and lays a solid foundation for genetic improvement of flowering cherry germplasm.

Methods

Genome assembly

For PacBio SMRT sequencing, the PacBio Sequel II platform was first used to generate sub-reads, and the sub-reads were then filtered by the ccs software using the parameter “min-passes=3, min-rq=0.99” to obtain 27.50 Gb of Hifi reads. The PacBio HiFi reads were assembled into the initial set of contigs using hifiasm v0.8¹⁶ with default parameters. The contig assembly had a total size of ~282.6 Mb, with a contig N50 value of 12.04 Mb. To achieve chromosome-level assembly, the ALLHiC algorithm was used to group, adjust the order and orientation of contigs and anchor the assembled contigs into eight pseudomolecules based on Hi-C data. After ALLHiC scaffolding, Hi-C interaction heat map was constructed using HiC-Pro v3.1.0 and visualized using HiCPlotter.

Genome annotations

For repeat sequence annotation, we used a combined strategy of homology-based search and de novo prediction. The homology-based search was based on the Repbase database using RepeatMasker (http://www.repeatmasker.org/) and RepeatProteinMask (http://www.repeatmasker.org/) to search for interspersed repeat elements. The de novo prediction was based on a species-specific repeat database generated by LTR_FINDER, Piler , RepeatScout and RepeatModeler. Using this library, we identified de novo involved repeats with RepeatMasker and predicted tandem repeats with TRF. In addition, we used RepeatMasker to mask the repetitive sequences as input for gene structure prediction.

For protein-coding gene structure prediction, we used a comprehensive approach that integrates homology-based prediction, de novo prediction, and RNA-Seq-based prediction. For homology-based prediction, Blast and Genewise were used to align the amino acid sequences from the Malus domestica, P. serrulata, Vitis vinifera, A. thaliana, and P. salicina genomes to the assembled P. campanulata sequences. Augustus, GlimmerHMM, SNAP, GeneID, and GenScan were used to predict de novo gene models. Cufflinks and PASA were applied to predict the gene models in the RNA-Seq-based prediction study. The results of the above three approaches were further integrated to generate a final non-redundant gene model set using EVidenceModeler and modified using PASA.