*2.8. Genotyping*

Using the C7AIR, 7098 SNPs were called for N22, Swarna and the DTY-IL [58]. Frequencies for each SNP across replicated samples were estimated and the most frequent genotype was considered true. Missing markers and monomorphic markers between the parents N22 and Swarna were discarded. For the remaining markers (1648 SNPs) genotypic calls from each parent were used to assign inheritance from N22 or Swarna. Fragments from the donor parent N22 were defined as consecutive SNPs with N22 homozygous genotypes. Markers that potentially represent miss-called double recombination were discarded if the probability of observing this event was smaller than 1 cM or 1 in 100 events. A graphical representation of the markers inherited from N22 and Swarna were graphed using the R package ggplot2. RNA-Seq reads of the candidate genes were aligned and visualized in IGV (version 2.8.2) relative to MSU7 (Nipponbare) and in the latest versions of the MH63 and N22. Clustal Omega (version 1.2.4) was used for multiple protein sequence alignments. FGENESH was used as the gene prediction tool with *O. sativa* vg. *indica* as the background organism. SNP-Seek (https://snp-seek.irri.org) was used to validate the NS SNPs across the 3 K genomes.
