Supplementary materials for PMGA

Plant mitochondrial genomes (PMGs) have many unique characteristics, including highly diverse genome structure, un-conserved inter-genetic regions, extensive RNA editing, and numerous horizontally transferred sequences. PMG annotators aware of these characteristics are missing, leading to inaccurate and incomplete annotation results. Here, we built a computational pipeline, "Plant Mitogenome Annotator" (PMGA), to address these challenges. PMGA had four datasets. Dataset 1 contained the coding sequences of rRNA and protein-coding genes (PCGs) of ten species. Erroneous splicing sites were corrected with RNA-seq data. Dataset 2 contained coding and protein sequences of 329 PMG. Erroneous annotations were corrected based on multiple sequence alignment. Dataset 3 contained 423 organelle’s tRNA sequences with defined origins. Dataset 4 contained plastid sequences from CPGAVAS2. In addition to annotate PCGs, rRNA genes, tRNA genes, PMGA can annotate MTPT sequences, repetitive sequences, chimeric genes, and RNA-editing sites. In particular, we developed three algorithms: "Analysis with Upstream Extended Sequences," "Assembling Exons with Weighted Direct Graph," and " Multiple Dimension Annotation of tRNA Genes" to annotate small exons, trans-splicing genes, and tRNA genes, respectively. We compared PMGA with three tools using the mitogenomes of Arabidopsis thaliana and Liriodendron tulipifera, we found that PMGA performed the best. Lastly, we sequenced, assembled and annotated the mitogenomes of Pueraria montana var. lobata and Eucommia ulmoides. We used the RNA-seq data, PCR, and Sanger sequencing results to determine the splicing sites. Comparison of the experimental and annotation results of the splicing sites showed 100% of accuracy. An online version of PMGA can be found at http://www.1kmpg.cn/pmga/.