Supplementary Figure and Tables

Supplementary figure and tables

Figure S1Marker-trait associations across genes in the vicinity of PHYC.

Table S1 Linkage mapping families

TableS2Primers and PCR conditions used to amplify sequences of the PHYC region for association mapping and signatures of selection studies

Table S3 BLASTN results of pearl millet sequences

Table S4 Comparison of results of the mixed model at PHYC and Pg7830

Table S5Neutrality tests applied to genes in the PHYC region in cultivated and wild pearl millet

Table S6 Effects estimates in the the inbred panel data*

Table S7 Effects estimates in linkage mapping data*

* Tables S6 and S7 are given as independent Excel files.

Figure S1Marker-trait associations across genes in the vicinity of PHYC.Associationswere fitted on the panel of inbreds using a mixed model. The p-value of the marker effect (Wald test) is given at each position (P-value was decimal log transformed). Trait and gene codes are given at the top of each graph. The horizontal line represents the 0.05 p-value threshold.

Table S1 Linkage mapping families

The code of the two parents, the type (wild/cultivated) and the country of origin is given for each family.

Table S2.Primers and PCR conditions used to amplify sequences of the PHYC region for association mapping and signatures of selection studies.

Gene name, forward and reverse primer sequences and melting temperature (Tm) for polymerase chain reaction (PCR) amplification for each of the genes and fragment size are presented. The data from these genes have been used for the detection of selection analysis (S), association mapping analysis (A) or both analyses.

Table S3 BLASTN results for pearl millet and sorghum sequences (or: Comparison of BLASTN results in pearl millet and sorghum sequences)

E-value of top BLASTN hits in pearl millet sequenced fragments and sorghum genes. The size of the sequenced fragments of pearl millet genes is given. We named the novel sequences of pearl millet using the corresponding gene index of the sorghum homolog preceded by the prefix Pg which refers to Pennisetum glaucum. Sorghum genes were named according to GRAMENE nomenclature (species name, chromosome index, gene index). Annotation corresponds to the annotation of the sorghum gene (the sign “-“ isdenotes uncharacherized protein). The gene-ontology (GO) of the biological process category is given.

Table S4 Comparison of results of the mixed model at PHYC and Pg7830

A mixed model was fitted to link polymorphisms with the flowering phenotype scored in 9 trials. The results of one marker each at PHYC and Pg7830 are compared with respect to the method used to handle missing sequence data. In the available dataset, entries with missing data differed between markers. In the first case, the analysis was performed after filtering entries with missing data for each marker on its own. In the second case, the dataset was filtered to choose only entries with valid sequence data for both markers (using the same subset). The number N of inbred in the final subset, the effect estimate (coefficient ±standard error, in days), the p-value for marker effect significance (Wald test) and the log-likelihood (LK) of the model are given. Based on different subsets, the Pg7830 marker yielded the highest level of significance and a higher effect estimate. Based on the same subset of inbreds, the situation was reversed and the PHYC marker yielded the highest level of significance and a higher effect estimate.

Table S5 Neutrality tests applied to the genes in the PHYC region in cultivated and wild pearl millet

For each gene, nucleotide diversity (π) for wild and cultivated samples, nucleotide polymorphism (θ) for wild (θw) and cultivated samples (θc), Tajima’s D for wild and cultivated samples, Fay and Wu’s H for wild and cultivated samples, and differentiation (FST) between wild and cultivated sample are shown. For each statistic (Tajima's D , Fay and Wu's H, FST) a p-value (P) is given based on a model of pearl millet domestication. P corresponds to the rank of the observed statistic value with respect to the expected distribution obtained by the EGGF model, divided by the number of simulations. The number of simulations (n) used to calculate the rank of FST for a given gene is shown. Simulations that shared similar θw per gene ± 0.1 were used. For Tajima’s D and Fay and Wu’s H, bilateral tests were used: * P<0.025 or P>0.975, *** P<0.0005 or P>0.9995. For FST, a unilateral test was used: * P>0.95. NA: outgroup sequence not available. Data from PHYC6b, PHYC6bF1, PHYC7, PHYC9, PHYC10, PHYC4 are from Clotault et al. (2012).