Computational protocol: A highly recombined, high‐density, eight‐founder wheat MAGIC map reveals extensive segregation distortion and genomic locations of introgression segments

Similar protocols

Protocol publication

[…] The MAGIC map was constructed in two steps using the R package mpMap version 1.25 (Huang and George, ) available from github (https://github.com/behuang/mpMap). For the first round of mapping, we used a subset of 18 750 SNP markers scored as co‐dominant. All heterozygote calls were set to missing. These markers were then filtered for missing founder genotypes as well as SD with P‐values <1e‐5. Recombination fractions between all pairs of markers were calculated using the function ‘mpestrf’ at default values. Markers were grouped hierarchically using the ‘mpgroup’ function into 300 linkage groups. These linkage groups were merged using R/mpMapInteractive (https://github.com/rohan-shah/mpMapInteractive) to produce larger groups which could then be assigned chromosome names based on marker groupings in previous genetic maps (Cavanagh et al., ; Wang et al., ; Genomezipper v5 (http://wheat-urgi.versailles.inra.fr/Seq-Repository/Genes-annotations) and in the Kansas deletion lines genotyped for the 80K array by Bristol University (http://www.cerealsdb.uk.net), as well as top BLASTn (Altschul et al., ) hits to IWGSC contigs listed in Wang et al. (, Table S6).Within linkage groups, markers were ordered using two‐point ordering implemented in the function ‘mporder’ using default settings. Fine ordering was performed interactively using R/mpMapInteractive. Map distances were computed using the Haldane mapping function using ‘computemap’. Once a draft map was built, previously excluded markers, including all those scored as dominant, were mapped as traits to the existing draft chromosomes, that is we treated the unmapped markers as phenotypes in a QTL analysis, similarly to Rostoks et al. (). Founder haplotype probabilities were computed with the ‘mpprob’ function in mpMap implemented in R/qtl (Broman et al., ) with a threshold of 0.6 and QTL were calculated using single‐stage QTL mapping in ‘mpIM’. Linkage groups were then reordered following the same steps as previously but including all the previously excluded loci which could be mapped to a chromosome with –log10 P > 16. In the construction of this final map, manual curation was used throughout the process and loci which could not be cleanly fitted into the ordered chromosomes were dropped. [...] Flanking marker sequences for both NIAB2105 and CM2014 were aligned to the 3B pseudomolecule (Choulet et al., ; https://urgi.versailles.inra.fr/download/wheat/3B/) using BLASTn. We only included alignments which were part of the final 3B pseudomolecule and had no mismatches. The strong Robigus SD locus was controlled for using the method described in Shah et al. (), and map distances were re‐estimated using the Haldane mapping function. This 783‐marker long MAGIC chromosome 3B had a reduced length of 230 cM compared to the original of 284 cM with 1408 markers. The number of crossovers per line was calculated using the function ‘mpprob’ in mpMap with either the options program = ‘qtl” (Broman et al., ) or program = ‘happy’ (Mott et al., ) and a threshold of 0.5. In R/qtl the number of crossovers was calculated using the function ‘calculateXO’ on default settings for eight‐parent RILs. Recombination per physical distance (cM/Mb) was calculated and visualized using R/MareyMap 1.3.0 (Rezvoy et al., ). A cubic spline with a smoothing parameter of spar 0.65 was fitted to calculate local recombination rates. Markers with inconsistent alignments were excluded from the fitting. Chromosome 3B of CM2014 was aligned to the 3B pseudomolecule as described above (with a total of 1406 markers), and its genetic distance rescaled to match the reduced 3B MAGIC map length of 230 cM. […]

Pipeline specifications

Software tools BLASTN, R/qtl, MareyMap
Databases CerealsDB
Application WGS analysis
Organisms Triticum aestivum