MEETING THE CHALLENGES OF NON-REFERENCED GENOME ASSEMBLY FROM SHORT-READ SEQUENCE DATA

Abstract

Massively parallel sequencing technologies (MPST) offer unprecedented opportunities for novel sequencing projects. MPST, while offering tremendous sequencing capacity, are typically most effective in resequencing projects (as opposed to the sequencing of novel genomes) due to the fact that sequence is returned in relatively short reads. Nonetheless, there is great interest in applying MPST to genome sequencing in non-model organisms. We have developed a bioinformatics pipeline to assemble short read sequence data into nearly complete chloroplast genomes using a combination of de novo and reference-guided assembly, while decreasing reliance on a reference genome. Initially, short read sequences are assembled into larger contigs using de novo assembly. De novo contigs are then aligned to the corresponding reference genome of the most closely related taxon available and merged to form a consensus sequence. The consensus sequence and reference are in turn 'merged' such that aligned de novo sequence remains unaffected while missing sequence is filled in using the reference sequence. This chimeric reference is then utilized in reference-guided assembly to align the original short read data, resulting in a draft plastome. Using two established Pinus reference plastomes, our method has been effective in the assembly of 33 chloroplast genomes within the genus Pinus, and results with four species representing other genera of Pinaceae suggest the method will be of general use in land plants, particularly once limitations of PCR-based chloroplast enrichment are overcome.