DRJs and breakpoint analysis in H. didymator

The proviral integrated segments are circularized and excised by homologous recombination between its extreme DRJs (at left and right extremities of the given segment, named DRJL and DRJR respectively). When the two copies of the DRJ exhibit some punctual differences, the excision site or breakpoint can be identified in a given recombined DRJ sequence with more or less resolution depending on the level of divergence between the DRJ copies.

In order to identify and analyze DRJ excision sites of a set of circularized IV sequences in an automatic fashion, we developed the following method called DrjBreakpointFinder and freely distributed at http:// github.com/stephanierobin/DrjBreakpointFinder/. The method takes as input a set of circularized sequences (usually obtained by sequencing) and a reference genome. It is composed of two main steps. The first step consists in identifying triplets of sequences (read-DRJL-DRJR) representing the recombined DRJ and its two parental DRJs, by mapping the sequencing reads to the reference genome. In the second step, a precise multiple alignment is computed for each sequence triplet, and a segmentation algorithm, inspired from the breakpoint refinement method Cassis [93], is applied along the recombined DRJ sequence to identify in the best case scenario the excision site or more generally the breakpoint region. To do so, the segmentation algorithm estimates the best partition of the recombined DRJ sequence into three distinct segments, corresponding to homology with DRJR, the breakpoint region, and homology with DRJL respectively, given the repartition of punctual differences with the two parental DRJs. The segmentation algorithm is classically based on fitting a piecewise constant function with two changepoints to the punctual difference signal (see [94]). DrjBreakpointFinder further gathers breakpoint results by proviral segments or DRJ pairs, in order to obtain for each the distribution of potential excision sites observed in a given circular virus sequencing dataset. The output of DrjBreakpointFinder consists of breakpoint region coordinate files along with visual representations for each proviral segment or DRJ pair.

In this paper, DrjBreakpointFinder was applied to two circular viral DNA sequencing datasets. Circular DNA was extracted from HdIV particles and sequenced by 454 and Sanger technologies, resulting in 40,343 and 15,575 reads, respectively [92].

In addition, the DRJ copy was manually analyzed for a subset of 8 segments (Hd12, Hd16, Hd19, Hd22, Hd24, Hd28, Hd29, and Hd30) that presented only one right and left DRJs in their integrated form. Junctions were amplified by PCR using primers located within the viral sequence, downstream and upstream the DRJs. PCR products were cloned in pGEM and 3 to 5 plasmid clones were then sequenced using Sanger technology for each segment. The obtained recombined junction sequences were then aligned with the 2 parental DRJs in an attempt to localize the excision site, based on the nucleotides differing between the 2 DRJs (see Additional file 11).