Assembly

      This assembly was based on 93.59 Gb Pacbio HiFi reads (Generated by the Sequel II sequencer (PacBio, USA) for 1800-minute movies each by Frasergen Bioinformatics Co., Ltd. (Wuhan, China)). The HiFi reads were assembled into contigs using Hifiasm v0.14-r312, and then purged using PurgeDups v1.2.3 4 and minimap2 v2.17-r941. The final assembly was scaffolded into chromosomes using 208.62 Gb Hi-C data (MGISEQ-2000) with juicer pipeline.

Annotation

       The gene annotation was performed using the MAKER v3.01.02 pipeline, combining evidence-based and ab initio approaches. Repeat masking was done using two libraries: a repeat library of Viridiplantae and a de novo library constructed with RepeatModeler2. Transcriptomic evidence was generated from Iso-Seq and RNA-Seq data and protein evidence was from protein sequences of Uniprot-sprot (taxonomy: viridiplantae), Arabidopsis thaliana (TAIR10), and Medicago truncatula (MtrunA17r5.0) Then, three rounds of gene prediction training were conducted, primarily to train de novo gene predictor models. The first round used transcription and homologous evidence to predict genes and train Augustus and SNAP. The second round used trained models of Augustus and SNAP with est2genome and protein2genome set to 0. The retrained models of SNAP and Augustus were used for the final round of gene annotation. Finally, gene structures were polished using PASA with high-quality FLNC reads.

Download

Fasta file	Polygala_tatarinowii.fa
GFF3 file	Polygala_tatarinowii.gff3
CDS file	Polygala_tatarinowii.cds
Protein file	Polygala_tatarinowii.pep