Rapid De Novo Assembly and Visualization of Plant and Animal Genomes Using Single-Library, PCR-Free, 250-Base Illumina Reads

Generation of high quality genome sequences for plant and animal species remains a fundamental challenge for the field, and this is particularly the case for larger genomes. Many models have been described for solving this problem, but they tend to be complicated and expensive.

Here we describe a low-cost and straightforward solution, based on a single PCR-free library coupled with a fast (24h on a single server) new algorithm, DISCOVAR de novo.  Sequencing was done using a HiSeq-2500 at 50x coverage with 2x250 base reads – now an officially supported protocol. For human this yielded 125 kb contigs - five times greater than obtained from ALLPATHS-LG using several libraries. Performance on other tested mammals (rhinoceros and aardvark) is also better.

DISCOVAR de novo assemblies are easily visualized graphs that capture polymorphism. Typically most of the genome is captured in long subgraphs that look like lines containing bubbles at heterozygous sites, and can be trivially flattened into standard FASTA contigs (with loss of information). By being completely friendly to polymorphism, DISCOVAR de novo thus solves a longstanding and vexing problem of the field.

While mammalian assemblies will have high contiguity, to get long scaffolds, one needs to add other data. We demonstrate that this can be done using BioNano Genomics maps. Clearly for particularly difficult genomes, such as crops, additional data will be needed to obtain a highly contiguous assembly. A workable approach is to start with a DISCOVAR de novo assembly graph, then use other data types to improve it.