Mitogenome assembly from genomic multiplex libraries: comparison of strategies and novel mitogenomes for five species of frogs

Abstract

Next-generation sequencing continues to revolutionize biodiversity studies by generating unprecedented amounts of DNA sequence data for comparative genomic analysis. However, these data are produced as millions or billions of short reads of variable quality that cannot be directly applied in comparative analyses, creating a demand for methods to facilitate assembly. We optimized an in silico strategy to efficiently reconstruct high-quality mitochondrial genomes directly from genomic reads. We tested this strategy using sequences from five species of frogs: Hylodes meridionalis (Hylodidae), Hyloxalus yasuni (Dendrobatidae), Pristimantis fenestratus (Craugastoridae), and Melanophryniscus simplex and Rhinella sp. (Bufonidae). These are the first mitogenomes published for these species, the genera Hylodes, Hyloxalus, Pristimantis, Melanophryniscus and Rhinella, and the families Craugastoridae and Hylodidae. Sequences were generated using only half of one lane of a standard Illumina HiqSeq 2000 flow cell, resulting in fewer than eight million reads. We analysed the reads of Hylodes meridionalis using three different assembly strategies: (1) reference-based (using bowtie2); (2) de novo (using abyss, soapdenovo2 and velvet); and (3) baiting and iterative mapping (using mira and mitobim). Mitogenomes were assembled exclusively with strategy 3, which we employed to assemble the remaining mitogenomes. Annotations were performed with mitos and confirmed by comparison with published amphibian mitochondria. In most cases, we recovered all 13 coding genes, 22 tRNAs, and two ribosomal subunit genes, with minor gene rearrangements. Our results show that few raw reads can be sufficient to generate high-quality scaffolds, making any Illumina machine run using genomic multiplex libraries a potential source of data for organelle assemblies as by-catch.