The plastid genome (plastome, plastids including the chloroplast and other plastid forms) and mitochondrial genome (mitogenome or chondriome) represent the portions of endosymbiotic organelle inheritance in eukaryotes that have remained in organelles without being transferred to the nucleus or lost.

The DNA sequenes from the organelle genomes have been widely used in phylogenetic and evolutionary analyses, and DNA barcoding. Due to high copy numbers of the organelle genome in a single cell, it is feasible to get enough coverage from the low coverage whole genome sequencing (WGS) data to assemble complete organelle genomes.

With the rapid advances of high throughput sequencing technologies, a tremendous amount of WGS data were produced in low cost, which makes the accurate and high throughput assembly of organelle genomes in great need. Although many toolkits and pipelines for assembling organelle genomes have been developed, their assembly qualities and efficiencies are generally below expectations.

Teams led Prof. LI Dezhu and Prof. YI Tingshuang from Kunming Institute of Botany, Chinese Academy of Sciences (KIB/CAS) have been engaged in plastid phylogenomics, comparative genomics, and DNA barcoding for years.

Jointly, they have established a research system of utilizing the plastome data for phylogenetic and evolutionary studies, and achieved fruitful results (Ma et al., 2014. Systematic Biology；Zhang et al., 2017. New Phytologist；Li et al., 2019. Nature Plants；Zhang et al.,2020 Systematic Biology).

Teams were also dedicated to develop new tool for plastome analyses, including a popular plastome annotation toolkit, PGA (Qu et al., 2019, Plant Methods).

Aiming at accurate and efficient organelle genome assembly, an international joint team led by Profs. Li and Yi, with collaborators from KIB and , Xishuangbanna Tropical Botanical Gardens, Chinese Academy of Sciences (XTBG/CAS), and Pennsylvania State University, newly developed GetOrganelle, an advanced toolkit for de novo assembling accurate organelle genomes.

The “from-reads-to-organelle” process, i.e. the main workflow of GetOrganelle can be summarized into the following five steps:

1) recruiting initial “seed reads” and estimating parameter,

2) recruiting more target-associated reads through extending,

3) conducting de novo assembly and generating assembly graph,

4) roughly filtering for target-like contigs,

5) identifying target contigs and exporting all configurations.

Figure 1:  The workflow of GetOrganelle toolkit. (Image by KIB)

It is innovative that GetOrganelle provides a pre-grouping strategy for speeding up target-associated reads recruitment and contig multiplicity estimation algorithm for better repeat resolution.

The new algorithm of contig multiplicity estimation incorporates both information of graph characteristics and contig coverage (Fig. 2).

Figure 2:  An example of exporting all configurations from a target-complete assembly graph in GetOrganelle. (Image by KIB)

To evaluate the accuracy of assemblies using GetOrganelle, in comparison with another popular assembler NOVOPlasty, the GetOrganelle team tested the 156 public datasets from plants, animals, and fungi. For 50 public plant WGS datasets, GetOrganelle generated a high completeness rate of 78% with default settings, significantly better than currently most popular tool NOVOPlasty which generated 16% with fine-tuning but cost slightly less computational resources.

GetOrganelle still significantly outperformed NOVOPlasty at completeness rate even when consuming comparable or less computational resources. Furthermore, NOVOPlasty generated 20%~25% wrong/false complete plastomes in K=23 and K=31 runs (Fig. 3).

In the same test, the consistency of GetOrganelle assemblies under different parameters was also better than that of NOVOPlasty assemblies.