An Improved Assembly Algorithm for de novo Circular Genome Reconstruction

Circular chromosomes or genomes, such as viruses, bacteria, mitochondria and plasmids, are a common occurrence in nature, but despite the wide array of algorithms available for de novo assembly, the circularity of these DNA molecules is largely overlooked. There are some limitations of this oversight. Current NGS assembly algorithms assume a linear molecule and will result in a linearly represented genome, with a breakpoint in an arbitrary position. This is becoming increasingly problematic with increasing NGS read lengths resulting in fewer contig assemblies with less ambiguity. Additionally, long read sequencing technologies promise to provide sequences that may greatly span breakpoints at the expense of coverage - resulting in a relatively large quanta of information loss.

Although there are currently methods for the re-circularisation of contigs post-assembly by identifying common trailing/leading sequence motifs, we present a more robust approach of circularising during the assembly process whilst still allowing the merging of similar and sub-contigs throughout the overlap-based approach. This method can also combat the issue of chimeric sequence due to contamination, a common problem when sequencing bacterial cultures, due to decreased likelihood of conserved contig ends due to timely circularisation.

We present results from both a simulated dataset represented by a 3,000 read library generated from a published 7 Kbp human papillomavirus (HPV) genome, and a real-world dataset using a 267,491 read Panthera leo persica(Asiatic Lion) mitochondrial NGS library produced using an Ion Torrent sequencing machine. These results are discussed and compared to some of the more popular linear assembly algorithms in common usage today.

Geneious R8 is the first bioinformatics software package to offer a circular de novo assembly method. The Geneious Circular de novo assembler is developed by Biomatters and may be found at http://www.geneious.com