Abstract
MotivationIntimately tied to assembly quality is the complexity of the de Bruijn graph built by the assembler. Thus, there have been many paradigms developed to decrease the complexity of the de Bruijn graph. One obvious combinatorial paradigm for this is to allow the value of k to vary; having a larger value of k where the graph is more complex and a smaller value of k where the graph would likely contain fewer spurious edges and vertices. One open problem that affects the practicality of this method is how to predict the value of k prior to building the de Bruijn graph. We show that optimal values of k can be predicted prior to assembly by using the information contained in a phylogenetically-close genome and therefore, help make the use of multiple values of k practical for genome assembly.ResultsWe present HyDA-Vista, which is a genome assembler that uses homology information to choose a value of k for each read prior to the de Bruijn graph construction. The chosen k is optimal if there are no sequencing errors and the coverage is sufficient. Fundamental to our method is the construction of the maximal sequence landscape, which is a data structure that stores for each position in the input string, the largest repeated substring containing that position. In particular, we show the maximal sequence landscape can be constructed in O(n + n log n)-time and O(n)-space. HyDA-Vista first constructs the maximal sequence landscape for a homologous genome. The reads are then aligned to this reference genome, and values of k are assigned to each read using the maximal sequence landscape and the alignments. Eventually, all the reads are assembled by an iterative de Bruijn graph construction method. Our results and comparison to other assemblers demonstrate that HyDA-Vista achieves the best assembly of E. coli before repeat resolution or scaffolding.AvailabilityHyDA-Vista is freely available [1]. The code for constructing the maximal sequence landscape and choosing the optimal value of k for each read is also separately available on the website and could be incorporated into any genome assembler.
Highlights
The ability to accurately assemble genomes is a fundamental problem in bioinformatics that is vital to the success of many scientific projects, including the 10,000 vertebrate genomes (Genome 10K) [2], Arabidopsis variations (1001 genomes) [3], human variations (1000 genomes) [4], and Human Microbiome Project [5]
The reads are aligned to this reference genome, and values of k are assigned to each read using the maximal sequence landscape and the alignments
Algorithm 1 An overview of HyDA-Vista 1: Build the maximal sequence landscape for the reference genome. 2: Align all reads to the reference using BWA. 3: For each aligned read: assign a value of k using the maximal sequence landscape. 4: Unaligned reads are assigned a value of k using a heuristic. 5: The de Bruijn graph is constructed in an iterative manner, as shown in Algorithm 2
Summary
The ability to accurately assemble genomes is a fundamental problem in bioinformatics that is vital to the success of many scientific projects, including the 10,000 vertebrate genomes (Genome 10K) [2], Arabidopsis variations (1001 genomes) [3], human variations (1000 genomes) [4], and Human Microbiome Project [5]. In Eulerian sequence assembly [9,10], a de Bruijn graph is constructed with a vertex v for every (k − 1)-mer present in an input set of reads, and an edge v − v for every observed k-mer in the reads with (k − 1)-mer prefix v and (k − 1)-mer suffix v. The majority of de Bruijn graph based assemblers follow the same general outline: break the (possibly error corrected) reads into k-mers, construct the de Bruijn graph on the set of resulting k-mers, simplify the de Bruijn graph, resolve the repeated regions by using mate-pair information, and construct the contigs (simple paths in the de Bruijn graph). We use sij , where 1 ≤ i ≤ j ≤ n, to indicate substring sisi+1 sj of string s.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
