WhatsHap: Weighted Haplotype Assembly for Future-Generation Sequencing Reads

Abstract

The human genome is diploid, which requires assigning heterozygous single nucleotide polymorphisms (SNPs) to the two copies of the genome. The resulting haplotypes, lists of SNPs belonging to each copy, are crucial for downstream analyses in population genetics. Currently, statistical approaches, which are oblivious to direct read information, constitute the state-of-the-art. Haplotype assembly, which addresses phasing directly from sequencing reads, suffers from the fact that sequencing reads of the current generation are too short to serve the purposes of genome-wide phasing. While future-technology sequencing reads will contain sufficient amounts of SNPs per read for phasing, they are also likely to suffer from higher sequencing error rates. Currently, no haplotype assembly approaches exist that allow for taking both increasing read length and sequencing error information into account. Here, we suggest WhatsHap, the first approach that yields provably optimal solutions to the weighted minimum error correction problem in runtime linear in the number of SNPs. WhatsHap is a fixed parameter tractable (FPT) approach with coverage as the parameter. We demonstrate that WhatsHap can handle datasets of coverage up to 203, and that 153 are generally enough for reliably phasing long reads, even at significantly elevated sequencing error rates. We also find that the switch and flip error rates of the haplotypes we output are favorable when comparing them with state-of-the-art statistical phasers.

HapCol: accurate and memory-efficient haplotype assembly from long reads

“…Both tools have been executed on all the instances, but HAPCOL terminated on some of them because no feasible solution existed for that choice of the input parameters. WHATSHAP, which should be able to find a feasible solution for all the instances, computed a solution only for the instances with coverage 15Â and 20Â, while, as expected (Patterson et al, 2015), it was not able to successfully conclude the execution on the instances with coverage 25Â since it exhausted the available memory (256 GB). Table 2 reports, for any combination of input parameters and a, the number of instances with a feasible solution (column 'feas.…”

“…We compared HAPCOL with three state-of-the-art haplotyping tools specifically designed for handling long reads, namely, REFHAP, which was shown to be one of the most accurate heuristic methods (Duitama et al, 2012), PROBHAP, a recent probabilistic method which has been shown to be sensibly more accurate than REFHAP (Kuleshov, 2014) and WHATSHAP, the first exact approach for the weighted MEC problem specifically designed for long reads (Patterson et al, 2014(Patterson et al, , 2015. At higher coverages, applications such as SNP calling or validating which SNPs are really heterozygous in the given sample (e.g.…”

“…As motivated before, in this part we focused on the tools that can work also without the all-heterozygous assumption, namely HAPCOL and WHATSHAP. The simulation of the datasets has been performed as in Patterson et al (2015). The dataset consists of a ground truth, which was assembled by inserting all known variants of chromosomes 1 and 15 of J. Craig Venter's genome into the Accuracy is given in terms of phasing error ('error') and total phased positions ('phased') of the reconstructed haplotypes, while performances are given in terms of total running time ('time') expressed in seconds and the peak memory usage ('mem.')…”

“…Recently, MEC and wMEC approaches have been used in the context of long reads, confirming that long fragments allow to assemble haplotypes more accurately than traditional short reads (Aguiar and Istrail, 2012;Duitama et al, 2012;Patterson et al, 2014Patterson et al, , 2015. Since MEC is NP-hard (Cilibrasi et al, 2007), exact solutions have exponential complexity.…”

H-PoP and H-PoPG: heuristic partitioning algorithms for single individual haplotyping of polyploids

“…Most of the above models and their extended versions are NP-hard (Bafna et al, 2005;Cilibrasi et al, 2007;Duitama et al, 2010), and their exact algorithms run in time exponential in at least one input parameter (Bafna et al, 2005;He et al, 2010;Xie et al, 2010bXie et al, , 2008Wang et al, 2010;Bonizzoni et al, 2015;Patterson et al, 2015;Pirola et al, 2015). Therefore, a large number of heuristic algorithms have been designed to deal with the problem (Panconesi and Sozio, 2004;Wang et al, 2005;Genovese et al, 2008;Duitama et al, 2010;Xie et al, 2012).…”