Computational protocol: Mapping Quantitative Trait Loci Affecting Biochemical and Morphological Fruit Properties in Eggplant (Solanum melongena L.)

Similar protocols

Protocol publication

[…] Statistical analyses were performed using R software (R Development Core Team, ). A conventional analysis of variance was applied to estimate genotype and environment effects based on the linear model Yij = μ + gi + bj + eij, where μ, g, b, and e represent, respectively, the overall mean, the genotypic effect, the block effect and the error. Broad-sense heritability values were given by σG2/([σG2 + σE2]/n), where σG2 represented the genetic variance, σE2 the residual variance and n the number of blocks. Correlations between traits were estimated using the Spearman coefficient, and normality, kurtosis and skewness were assessed with the Shapiro-Wilks test (α = 0.05). Segregation was considered as transgressive when at least one F2 individual recorded a trait value higher or lower by at least two standard deviations than the higher or lower scoring parental line. QTL detection was based on the Barchi et al. () map, constituted of 415 markers (339 SNPs, 2 HRMs, 3 CAPSs, 11 RFLPs, 33 SSRs, and 27 COSII) and spanning 1390 cM. Putative QTL location was determined by both interval (Lander and Botstein, ) and MQM (Jansen, ; Jansen and Stam, ) mapping, as implemented in MapQTL v5 software (Van Ooijen, ). QTL were initially identified using interval mapping, after which one linked marker per putative QTL was treated as a co-factor in the approximate multiple QTL model. Co-factor selection and MQM analysis were repeated until no new QTL could be identified. LOD thresholds for declaring a QTL to be significant at the 5% genome-wide probability level were established empirically by applying 1000 permutations per trait (Churchill and Doerge, ). Additive and dominance genetic effects, as well as the percentage of the phenotypic variance explained (PVE) by each QTL were obtained from the final multiple QTL model. The program QTLNetwork 2.1 (Yang et al., ) was used to analyze each set of environment's data separately to identify epistasis, and was then extended across both environments to identify any QTL x environment interactions present. QTL effects were estimated on the basis of the Markov Chain Monte Carlo (MCMC) method. A type I error level of 0.05 was applied. The genome scan employed a 10 cM window and a 1 cM walk speed. Critical F values were obtained by 1000 permutations and a threshold of 0.05 was applied to assign significance to a QTL or to an epistatic effect. Individual QTL were prefixed by a trait abbreviation, followed by the relevant chromosome designation, and were suffixed as “a” or “b” where more than one QTL mapped to a single linkage group; ML or MT was added as a suffix where the QTL was expressed in a site-specific manner. Epistatic effects were indicated by adding “*” to the label of a major established QTL, while “ep” was added to a newly detected QTL. MapChart v2.1 software (Voorrips, ) was draw the resulting maps. Syntenic regions of the tomato genome (sequence build 2.50; were accessed to identify candidate genes co-localizing with the eggplant QTL. Initial searches were conducted using 20-kb sections and, for sections of interest, additional searches were performed using 10 kb sections. Putative tomato orthologs of the eggplant genes were identified by Blast search in the tomato gene indices at DFCI. […]

Pipeline specifications

Software tools MapQTL, QTLNetwork, JoinMap
Application WGS analysis
Organisms Solanum melongena, Homo sapiens
Chemicals Anthocyanins, Carbohydrates, Chlorogenic Acid, Saponins, Phenol