Genome assembly and assessment
As for the genome assembly, all of the subreads was corrected by Falcon v1.8.7, (https://github.com/falconry/falcon/releases) with specific parameters (length_cutoff = 18,000; length_cutoff_pr = 19,000) to generate the preads. And the initial genome was assembled with smartdenovo (wtpre -J 3000, wtzmo -k 21 -z 10 -Z 19 -U -1 -m 0.1 -A 1000, https://github.com/ruanjue/smartdenovo) by using the corrected preads. In order to produce more precise genome sequence, initial genome was polished by Arrow with all of subreads based on default parameters. All high-quality NGS data was used to polish the Arrow-correct genome by nextpolish with specific parameters (task=12121212) to obtain the polished genome (Walker, et al. 2014; Hu, et al. 2020). Finally, to acquire non-redundant haploid genome, some short and redundant sequences were removed from the polished genome by using redundans (Pryszcz and Gabaldon 2016) with some parameters (identity=0.824; coverage=0.8).
To assess the precise and non-redundant of genome, we carried out four methods as follows: (1) RNA-seq data were mapped to G. przewalskigenome by using hisat2 (Pertea, et al. 2016) with default parameters for the accuracy of gene regions (2) the genome of subreads data and NGS data were mapped to genome with minimap2 (-x pb) and bwa based on default parameters, respectively for the accuracy of assembly sequences (Li and Durbin 2009; Li 2018) . (3) the NGS mapping file was utilized to analysis the genome single-base accuracy by calling SNPs and Indels. (4) BUSCO database (https://busco.ezlab.org/) was employed to assess the completeness of genome with default parameters.