An Efficient Preprocessing Module for Joining Paired end Reads before Assembly

We came across this interesting NGS assembly-related poster, while searching for ‘google sparseheap’ online. It is being presented without comment. More analysis will come twelve months later, when they submit their paper, get it peer-reviewed, revised, peer-reviewed again, edited, proof-read and published :)

**A one-step method for de novo genome assembly using short paired-end sequence reads

**

David H. Silver and Itai Yanai

Department of Biology, Technion Israel Institute of Technology, Haifa, Israel

Genome sequencing is currently only possible by assembling short sequence fragments of DNA by identifying overlapping regions. To help in this computationally intensive task, large DNA regions can be sequenced from both ends, yielding paired-end reads. Currently however, paired-end information is incorporated as an additional step following the assembly of contiguous sequences (contigs). The underlying reason for this has been the notion that this extra information derived from paired-end reads contributes mostly to distinguishing repeat regions, rather than to the main task of assembling. Here, we demonstrate that in fact the paired-end information is of tremendous aid to the assembly process itself. Further, we present a mathematical model which provides an explanatory framework for the success of our algorithm, thus closing the big gap between the theory of de novo assembly and the practice. Our algorithm benefits de novo genome assembling by a better assembly in terms of sequence length with fewer assembly errors and requiring significantly less sequence coverage.