Minichromosome genomes reconstruction and comparison
Due to the conserved noncoding region shared by all minichromosomes, the Spades based assembly was not able to separate the different minichromosomes reliably. We, therefore, took an alternative approach. Using a single gene from each minichromosome as a reference (preliminary assignment of the genes to individual minichromosomes was based on the GenBank data available for P. spinulosa ; acc. nos. KF647762-KF647771) we mapped the reads and extended the sequences in the program aTRAM 2.0 (Allen at al., 2018). To obtain annotations of the resulting sequences, we combined two methods. The first method utilized the web based server Mitos (Bernt et al., 2013). Since this method missannotated or entirely missed some of the genes, we corrected the results by a blast based approach, taking advantage of the available annotated mitochondrial genome from P. spinulosa . We combined assembled minichromosome sequences into a custom database of P. serrata and used the 37 genes of P. spinulosa as blast queries (we run discontinous megablast and tblastx, both with E-value set to 10). We then aligned the P. serrata minichromosome sequences together with P. asiatica and P. spinulosa using Mauve (Darling, Mau, Blattner, & Perna, 2004). These alignments were used as a background for combining and correcting the annotations. To prepare a concatenated matrix, we trimmed all minichromosomes to equal lengths. Phylogenetic tree and genetic distances were retrieved from concatenated alignments by the same approach as for the L. polyplacis genomes.