Schematic view of the local haplotype reconstruction. From the multiple sequence alignment of all reads, a window is defined and reads overlapping it are extracted and passed (in fasta format) to the program diri_sampler. The program returns (1) the inferred haplotype sequences, (2) their frequencies, (3) the set of corrected reads and (4) the full posterior probability of the reconstruction (not reported in the Figure). Same colour indicates reads originating from the same haplotype. Dots represent errors on the reads that are corrected in order to infer the haplotypes (thicker lines).