trio-sga: facilitating de novo assembly of highly heterozygous genomes with parent-child trios

Motivation: Most DNA sequence in diploid organisms is found in two copies, one contributed by the mother and the other by the father. The high density of differences between the maternally and paternally contributed sequences (heterozygous sites) in some organisms makes de novo genome assembly very challenging, even for algorithms specifically designed to deal with these cases. Therefore, various approaches, most commonly inbreeding in the laboratory, are used to reduce heterozygosity in genomic data prior to assembly. However, many species are not amenable to these techniques. Results: We introduce trio-sga, a set of three algorithms designed to take advantage of motherfather-offspring trio sequencing to facilitate better quality genome assembly in organisms with moderate to high levels of heterozygosity. Two of the algorithms use haplotype phase information present in the trio data to eliminate the majority of heterozygous sites before the assembly commences. The third algorithm is designed to reduce sequencing costs by enabling the use of parents’ reads in the assembly of the genome of the offspring. We test these algorithms on a ‘simulated trio’ from four haploid datasets, and further demonstrate their performance by assembling three highly heterozygous Heliconius butterfly genomes. While the implementation of trio-sga is tuned towards Illuminagenerated data, we note that the trio approach to reducing heterozygosity is likely to have crossplatform utility for de novo assembly. Availability: trio-sga is an extension of the sga genome assembler. It is available at https://github.com/millanek/trio-sga, written in C++, and runs multithreaded on UNIXbased systems. Contact: [email protected], [email protected]