Principle of the PR-BSA/BSR method
The principle of the PR-BSA/BSR method is presented in Fig. 1, in which peanut plant architecture traits are used as an example. A segregating population, including F2 generation-derived recombinant inbred lines (RILs), was obtained from a cross between two inbred varieties or lines with substantial genetic polymorphisms. We selected independent target traits A and B, which differed significantly between the two parental lines (Fig. 1A). The segregating progenies were divided into the following two types according to the two traits: parent-type and recombinant-type. Specifically, in the parent-type progenies, both traits were derived from one parent, with the female and male parental traits designated as AfBf and AmBm, respectively. In contrast, in the recombinant-type progenies, one trait was derived from each parent. For example, trait A was from the male parent and trait B was from the female parent (AmBf). Alternatively, trait A was from the female parent and trait B was from the male parent (AfBm) (Fig. 1B).
We selected individuals with extreme traits from among the progenies with recombinant phenotypes to construct AmBf and AfBm bulked pools. The parents and the two bulks were analyzed by genome resequencing or transcriptome sequencing. We subsequently further analyzed the transcriptome sequencing data. Clean reads obtained from the parental sequencing data were compared with the reference genome to detect SNP sites. All SNP sites were compared and the ones that differed between the parents were selected to calculate the SNP-index. Therefore, there were two types of SNPs for genotyping: SNPM (same as the male parent) and SNPF (same as the female parent).
The parental SNP genotypes were determined and the proportion of each SNP among all reads in the two pools was calculated (i.e., SNPM-index and SNPF-index). Next, the ΔSNP-index was calculated by subtracting the SNPs between the two pools, which can also reveal the direction of the two traits in the graph and further identify the trait-related loci. For trait A, the D-value of the two pools was calculated as AmBf − AfBm (SNPF-index − SNPM-index), which resulted in a positive peak for the ΔSNP-index associated with trait A. In contrast, for trait B, the SNPM-index was lower in the AmBf pool than in the AfBm pool. Accordingly, when SNPM-index was subtracted from SNPF-index, the ΔSNP-index associated with trait B was detected as a negative peak (Fig. 1C).
Genetic Analysis Of Lba And Fbp
The LBA and FBP phenotypic segregation data for PF–F6 are provided in Tables 1. For LBA, 115 RILs exhibited a prostrating phenotype, with only the tips of the lateral branches curved upward (TLB), whereas 133 RILs exhibited a spreading phenotype, but the middle of the lateral branches curved upward (MLB), TLB and MLB both belong to the spreading growth type. A total of 202 RILs exhibited an erect phenotype. Hence, the spreading-type: erect-type segregation ratio was 248:202, which is approximately 1:1. For FBP, 256 and 211 RILs were designated as alternating-type and sequential-type, respectively. Thus, the alternating-type: sequential-type segregation ratio was approximately 1:1 (Table 1). Therefore, LBA and FBP are likely regulated by a major gene.
Mapping Of Loci Related To Lba And Fbp By Pr-bsa/bsr
We used the PR-BSA/BSR method to detect loci related to LBA and FBP in peanut. The F6 population derived from a cross between Pingdu9616 (Female parent) and Florunner (male parent) was used for gene mapping. Individuals with phenotypes that were similar to the parental phenotypes were considered as extreme-phenotype individuals. Moreover, the extreme-phenotype individuals of the F6 population were divided into four types (Fig. 1B). Specifically, there were two parent-types: Pingdu9616 was LBA-erect and FBP-sequential (LeFs) and Florunner was LBA-spreading and FBP-alternating (LsFa). Additionally, there were two recombinant phenotypes (PR pool), which were LBA-erect and FBP-alternating (LeFa) and LBA-spreading and FBP-sequential (LsFs). Two bulks were constructed from 30 LeFa individuals (EA bulk) and 30 LsFs individuals (SS bulk). Bulks were not prepared for the parent-types because the allelic variants in their genomes were from the same parents, making it impossible to directly distinguish the trait-related loci.
To eliminate false-positive sites, the SNP-index or ΔSNP-index of the markers on the same chromosome were fitted on the basis of the position of the marker on the genome according to the SNPNUM method (Liu et al. 2012). To clarify the relationship between the traits and the direction of the peaks, the SNP-index of the two pools was calculated, which reflected the proportion of the allelic variations from Pingdu9616 (SNPF) among all reads. For LBA, there were more SNPFs from the LeFa pool than from the LsFs pool. For FBP, there were fewer SNPFs from the LeFa pool than from the LrFs pool. When the SNPF-index (LsFs) was subtracted from the SNPF-index (LeFa), we detected a positive peak associated with LBA in the 6.82 Mb region (101,743,223–108,564,267 bp) on chromosome 15 as well as a negative peak associated with FBP in the 2.16 Mb region (117,682,534–119,846,824 bp) on chromosome 12 (Fig. 2).
Validation According To Individual Snp-indices
Transcriptome sequencing data for individuals and not mixed samples were analyzed in this study to determine the SNP genotypes of individuals. Therefore, the SNP-index of each mixed pool and the ΔSNP-index between the two pools were accurately calculated by confirming the SNP genotypes of a single individual. For example (Fig. 1C), assuming we determined the SNP genotypes of the female parent and there were five individuals in the SS/EA pool as well as two and three individuals with SNPF and SNPM, respectively, then SNPM-index = 2/5 and SNPF-index = 3/5. We recalculated the SNP-index and ΔSNP-index according to this method for analyzing individual SNP-indices and prepared a scatter diagram of chromosomes B05 and B02 to verify the PR-BSA/BSR results (Fig. 3). When the SNP-index of the EA pool was subtracted from the SNP-index of the SS pool, the resulting negative peak corresponded to the end of chromosome 15, which is consistent with the peak direction and chromosomal position revealed by the PR-BSA/BSR method. The positive peak corresponding to the end of chromosome 12 was also in accordance with the PR-BSA/BSR results. Thus, consistent results were obtained by the two algorithms, reflecting the accuracy and reliability of the PR-BSA/BSR method.
Validation By Linkage Mapping
We developed 31 and 20 pairs of InDel markers in the candidate regions of chromosome 15 (101,743,223–108,564,267 bp) and chromosome 12 (117,682,534–119,846,824 bp), respectively (Supplementary Table 1). A total of 445 families in the RIL–PF–F6 populations were used for genotyping and analyses. We constructed genetic linkage maps for chromosomes 15 and 12 to identify and verify the loci related to LBA and FBP, respectively.
Among the InDel markers on chromosome 15, 12 were polymorphic between the parents. A genetic linkage map was constructed, with a total length of 192.19 cM. a major QTL for LBA flanked by InDel markers J15-12 and J15-11 on chromosome 15 was identified (Fig. 4A). The QTL explained 44.59% of the phenotypic variance and had an LOD value of 27.05 (Table 2). This QTL mapping result was consistent with the PR-BSA/BSR result, suggestive of a major QTL associated with LBA at the end of chromosome 15.
Of the InDel markers on chromosome 12, seven co-dominant markers were polymorphic between the parents. A local genetic linkage map was constructed, with a total length of 26.81 cM. On the basis of linkage mapping, the QTL for FBP was located between markers GPR2-21 and PB2-29 at the end of chromosome B02 (Fig. 4B). The QTL explained 85.75% of the phenotypic variance and had an LOD value of 467.30 (Table 2).This result was consistent with the PR-BSA/BSR result.