Fig. 1

Comparative Subsequence Sets Analysis (CoSSA) workflow example. 1 The R-bulk, the S-bulk and (optionally) the parents of the full-sib population are sequenced with Illumina 151 bp PE. The Illumina reads are quality trimmed and k-mer tables are built for each of the samples. The k-mer tables contain all the possible k-mers of the sample and their frequency (~sequencing depth). K-mers with a frequency of 1 are considered as sequencing errors and are removed from the set. The difference between the R-bulk and S-bulk k-mer tables is performed to keep only the R-bulk specific k-mers. 2 R-bulk specific k-mers are filtered upon their depth in function of the expected depth of the R haplotype. In our situation, we worked with a tetraploid population in which the resistance segregated with a 1:1 segregation ratio. Each of the progeny clones pooled in the R-bulk held the R-haplotype in one copy (simplex) which means that 25% of the reads cover the R-haplotype. The depth of each haploid genome in our R-bulk was ~ 16 × so we kept the k-mers with a depth from 10 × to 22 ×. If the parents have been sequenced, additional set operations can be used to divide the k-mers in function of their parental origin: k-mers inherited from the resistant parent (Kuba, subset A: resistance specific k-mers), k-mers inherited from the susceptible parent (Ludmilla, subset B), k-mers inherited from both parents (subset C) and the k-mers inherited from none of the parents (subset D). 3a If a reference genome is available, the k-mers are mapped to it to identify the resistance locus. 3b If no reference genome is available, the read pairs containing the R-bulk specific k-mers are extracted from the raw reads data and used to re-assemble the resistant haplotype. Nu: Number of unique k-mers, Nt: Total number of k-mers