Limits...
A high-density, multi-parental SNP genetic map on apple validates a new mapping approach for outcrossing species

View Article: PubMed Central - PubMed

ABSTRACT

Quantitative trait loci (QTL) mapping approaches rely on the correct ordering of molecular markers along the chromosomes, which can be obtained from genetic linkage maps or a reference genome sequence. For apple (Malus domestica Borkh), the genome sequence v1 and v2 could not meet this need; therefore, a novel approach was devised to develop a dense genetic linkage map, providing the most reliable marker-loci order for the highest possible number of markers. The approach was based on four strategies: (i) the use of multiple full-sib families, (ii) the reduction of missing information through the use of HaploBlocks and alternative calling procedures for single-nucleotide polymorphism (SNP) markers, (iii) the construction of a single backcross-type data set including all families, and (iv) a two-step map generation procedure based on the sequential inclusion of markers. The map comprises 15 417 SNP markers, clustered in 3 K HaploBlock markers spanning 1 267 cM, with an average distance between adjacent markers of 0.37 cM and a maximum distance of 3.29 cM. Moreover, chromosome 5 was oriented according to its homoeologous chromosome 10. This map was useful to improve the apple genome sequence, design the Axiom Apple 480 K SNP array and perform multifamily-based QTL studies. Its collinearity with the genome sequences v1 and v3 are reported. To our knowledge, this is the shortest published SNP map in apple, while including the largest number of markers, families and individuals. This result validates our methodology, proving its value for the construction of integrated linkage maps for any outbreeding species.

No MeSH data available.


Graphical visualization of the combined HaploBlock and backcross approach presented in the current study. The figure illustrates the main steps of the process with an example from the true data of five families, each represented by seven individuals, two HaploBlocks (HBs) and one individual SNP on linkage group 1. Genotype codes presented here follow the format of JoinMap v3 and later versions for the cross-pollinated (CP) segregation types (Segr), where <lmxll> refers to a maternal marker with genotypes lm and ll, <nnxnp> to a paternal marker with genotypes nn and np, and <hkxhk> refers to a bi-parental marker with genotypes hh, hk and kk (see https://www.kyazma.nl/docs/JM4manual.pdf—Table 4). These three segregation types are highlighted with different colors: red for markers segregating only in the mother, blue for markers segregating only in the father, and green for those segregating in both parents; missing data (−−) and initially non-informative codes (hk) are not highlighted. (a) The use of the HB strategy allowed the identification of stable sets of SNP-markers, such as those composing HB_1430 and HB_902 that consist of 6 and 10 SNPs, respectively. These SNPs do not segregate in all families (the only exception is F_0420898_L1_PA), thus leading to a considerable amount of missing information (62% of data points). (b) The genotypic information of the co-segregating SNPs is aggregated to form a single HB marker across families and the bi-parental allelic contribution is also split to form two distinct single-parent data sets, where the phase of the new ‘single parent’ HB-markers is adjusted accordingly. (c) The two complete single-parent data sets are subsequently converted in a backcross (BC) design and combined to form a unique population of twice the number of individuals as the initial CP populations. The presented strategy permits the almost complete exploitation of the segregation information available (losing only some information from the rare recombination events within a HB) while considerably reducing the amount of missing information: in this example, from 76% for the initial CP data sets of the two HBs to 28% in the final unique BC population. For the single SNP, the amount of missing data did not change throughout the process by definition and was 66%. This approach of data aggregation and mating type was implemented in the software Haplotype Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm), whose manual describes the process in more detail.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5120355&req=5

fig1: Graphical visualization of the combined HaploBlock and backcross approach presented in the current study. The figure illustrates the main steps of the process with an example from the true data of five families, each represented by seven individuals, two HaploBlocks (HBs) and one individual SNP on linkage group 1. Genotype codes presented here follow the format of JoinMap v3 and later versions for the cross-pollinated (CP) segregation types (Segr), where <lmxll> refers to a maternal marker with genotypes lm and ll, <nnxnp> to a paternal marker with genotypes nn and np, and <hkxhk> refers to a bi-parental marker with genotypes hh, hk and kk (see https://www.kyazma.nl/docs/JM4manual.pdf—Table 4). These three segregation types are highlighted with different colors: red for markers segregating only in the mother, blue for markers segregating only in the father, and green for those segregating in both parents; missing data (−−) and initially non-informative codes (hk) are not highlighted. (a) The use of the HB strategy allowed the identification of stable sets of SNP-markers, such as those composing HB_1430 and HB_902 that consist of 6 and 10 SNPs, respectively. These SNPs do not segregate in all families (the only exception is F_0420898_L1_PA), thus leading to a considerable amount of missing information (62% of data points). (b) The genotypic information of the co-segregating SNPs is aggregated to form a single HB marker across families and the bi-parental allelic contribution is also split to form two distinct single-parent data sets, where the phase of the new ‘single parent’ HB-markers is adjusted accordingly. (c) The two complete single-parent data sets are subsequently converted in a backcross (BC) design and combined to form a unique population of twice the number of individuals as the initial CP populations. The presented strategy permits the almost complete exploitation of the segregation information available (losing only some information from the rare recombination events within a HB) while considerably reducing the amount of missing information: in this example, from 76% for the initial CP data sets of the two HBs to 28% in the final unique BC population. For the single SNP, the amount of missing data did not change throughout the process by definition and was 66%. This approach of data aggregation and mating type was implemented in the software Haplotype Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm), whose manual describes the process in more detail.

Mentions: The creation of HBs of co-segregating markers allowed a bin-mapping strategy where the segregation information of adjacent SNPs was aggregated and condensed into a single, virtual HB marker. The aggregation of co-segregating markers within the same HB increases the genotype score robustness consequently to information redundancy, and marker informativeness when combining markers with different segregation types. This is the case when a marker segregating in a single parent (for example, ab×aa) is combined with a bi-allelic marker heterozygous in both parental plants (for example, ab×ab) or with a single parent marker of the other sex (for example, aa×ab), leading to the generation of a fully informative marker record (corresponding to a segregation type ab×ac, or ab×cd). In view of our mapping effort, this strategy was implemented in the ad hoc developed software Haploblock Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm) and applied to our data (Supplementary File S1). For each FS family, HapAg aggregated the segregation information of the SNP markers belonging to the same HB by using the information on linkage group and the linkage phase of the individual markers (Figure 1a), while considering the meiotic events occurring in the two parental plants separately. Thereto, HapAg splits the parental allelic contribution of every individual of each family into two distinct sub-data sets including either maternal or paternal recombination events (Figure 1b). Eventually, maternal and paternal data sets of all the progenies from all FS-families were merged to generate a single BC-type data set (Figure 1c), having twice the number of individuals as the original CP populations (a more detailed description of the methodological steps performed by HapAg is available in the software manual (http://www.wageningenur.nl/en/show/HaploblockAggregator.htm)). The BC segregation type allows the correct phasing of the markers segregating in different families, leading to integrating the genotypic data prior to map construction and to the production of a unique integrated genetic map rather than a map resulting from the a posteriori integration of the linkage maps obtained from FS families.


A high-density, multi-parental SNP genetic map on apple validates a new mapping approach for outcrossing species
Graphical visualization of the combined HaploBlock and backcross approach presented in the current study. The figure illustrates the main steps of the process with an example from the true data of five families, each represented by seven individuals, two HaploBlocks (HBs) and one individual SNP on linkage group 1. Genotype codes presented here follow the format of JoinMap v3 and later versions for the cross-pollinated (CP) segregation types (Segr), where <lmxll> refers to a maternal marker with genotypes lm and ll, <nnxnp> to a paternal marker with genotypes nn and np, and <hkxhk> refers to a bi-parental marker with genotypes hh, hk and kk (see https://www.kyazma.nl/docs/JM4manual.pdf—Table 4). These three segregation types are highlighted with different colors: red for markers segregating only in the mother, blue for markers segregating only in the father, and green for those segregating in both parents; missing data (−−) and initially non-informative codes (hk) are not highlighted. (a) The use of the HB strategy allowed the identification of stable sets of SNP-markers, such as those composing HB_1430 and HB_902 that consist of 6 and 10 SNPs, respectively. These SNPs do not segregate in all families (the only exception is F_0420898_L1_PA), thus leading to a considerable amount of missing information (62% of data points). (b) The genotypic information of the co-segregating SNPs is aggregated to form a single HB marker across families and the bi-parental allelic contribution is also split to form two distinct single-parent data sets, where the phase of the new ‘single parent’ HB-markers is adjusted accordingly. (c) The two complete single-parent data sets are subsequently converted in a backcross (BC) design and combined to form a unique population of twice the number of individuals as the initial CP populations. The presented strategy permits the almost complete exploitation of the segregation information available (losing only some information from the rare recombination events within a HB) while considerably reducing the amount of missing information: in this example, from 76% for the initial CP data sets of the two HBs to 28% in the final unique BC population. For the single SNP, the amount of missing data did not change throughout the process by definition and was 66%. This approach of data aggregation and mating type was implemented in the software Haplotype Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm), whose manual describes the process in more detail.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5120355&req=5

fig1: Graphical visualization of the combined HaploBlock and backcross approach presented in the current study. The figure illustrates the main steps of the process with an example from the true data of five families, each represented by seven individuals, two HaploBlocks (HBs) and one individual SNP on linkage group 1. Genotype codes presented here follow the format of JoinMap v3 and later versions for the cross-pollinated (CP) segregation types (Segr), where <lmxll> refers to a maternal marker with genotypes lm and ll, <nnxnp> to a paternal marker with genotypes nn and np, and <hkxhk> refers to a bi-parental marker with genotypes hh, hk and kk (see https://www.kyazma.nl/docs/JM4manual.pdf—Table 4). These three segregation types are highlighted with different colors: red for markers segregating only in the mother, blue for markers segregating only in the father, and green for those segregating in both parents; missing data (−−) and initially non-informative codes (hk) are not highlighted. (a) The use of the HB strategy allowed the identification of stable sets of SNP-markers, such as those composing HB_1430 and HB_902 that consist of 6 and 10 SNPs, respectively. These SNPs do not segregate in all families (the only exception is F_0420898_L1_PA), thus leading to a considerable amount of missing information (62% of data points). (b) The genotypic information of the co-segregating SNPs is aggregated to form a single HB marker across families and the bi-parental allelic contribution is also split to form two distinct single-parent data sets, where the phase of the new ‘single parent’ HB-markers is adjusted accordingly. (c) The two complete single-parent data sets are subsequently converted in a backcross (BC) design and combined to form a unique population of twice the number of individuals as the initial CP populations. The presented strategy permits the almost complete exploitation of the segregation information available (losing only some information from the rare recombination events within a HB) while considerably reducing the amount of missing information: in this example, from 76% for the initial CP data sets of the two HBs to 28% in the final unique BC population. For the single SNP, the amount of missing data did not change throughout the process by definition and was 66%. This approach of data aggregation and mating type was implemented in the software Haplotype Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm), whose manual describes the process in more detail.
Mentions: The creation of HBs of co-segregating markers allowed a bin-mapping strategy where the segregation information of adjacent SNPs was aggregated and condensed into a single, virtual HB marker. The aggregation of co-segregating markers within the same HB increases the genotype score robustness consequently to information redundancy, and marker informativeness when combining markers with different segregation types. This is the case when a marker segregating in a single parent (for example, ab×aa) is combined with a bi-allelic marker heterozygous in both parental plants (for example, ab×ab) or with a single parent marker of the other sex (for example, aa×ab), leading to the generation of a fully informative marker record (corresponding to a segregation type ab×ac, or ab×cd). In view of our mapping effort, this strategy was implemented in the ad hoc developed software Haploblock Aggregator (HapAg—http://www.wageningenur.nl/en/show/HaploblockAggregator.htm) and applied to our data (Supplementary File S1). For each FS family, HapAg aggregated the segregation information of the SNP markers belonging to the same HB by using the information on linkage group and the linkage phase of the individual markers (Figure 1a), while considering the meiotic events occurring in the two parental plants separately. Thereto, HapAg splits the parental allelic contribution of every individual of each family into two distinct sub-data sets including either maternal or paternal recombination events (Figure 1b). Eventually, maternal and paternal data sets of all the progenies from all FS-families were merged to generate a single BC-type data set (Figure 1c), having twice the number of individuals as the original CP populations (a more detailed description of the methodological steps performed by HapAg is available in the software manual (http://www.wageningenur.nl/en/show/HaploblockAggregator.htm)). The BC segregation type allows the correct phasing of the markers segregating in different families, leading to integrating the genotypic data prior to map construction and to the production of a unique integrated genetic map rather than a map resulting from the a posteriori integration of the linkage maps obtained from FS families.

View Article: PubMed Central - PubMed

ABSTRACT

Quantitative trait loci (QTL) mapping approaches rely on the correct ordering of molecular markers along the chromosomes, which can be obtained from genetic linkage maps or a reference genome sequence. For apple (Malus domestica Borkh), the genome sequence v1 and v2 could not meet this need; therefore, a novel approach was devised to develop a dense genetic linkage map, providing the most reliable marker-loci order for the highest possible number of markers. The approach was based on four strategies: (i) the use of multiple full-sib families, (ii) the reduction of missing information through the use of HaploBlocks and alternative calling procedures for single-nucleotide polymorphism (SNP) markers, (iii) the construction of a single backcross-type data set including all families, and (iv) a two-step map generation procedure based on the sequential inclusion of markers. The map comprises 15&thinsp;417 SNP markers, clustered in 3&thinsp;K HaploBlock markers spanning 1&thinsp;267&thinsp;cM, with an average distance between adjacent markers of 0.37&thinsp;cM and a maximum distance of 3.29&thinsp;cM. Moreover, chromosome 5 was oriented according to its homoeologous chromosome 10. This map was useful to improve the apple genome sequence, design the Axiom Apple 480&thinsp;K SNP array and perform multifamily-based QTL studies. Its collinearity with the genome sequences v1 and v3 are reported. To our knowledge, this is the shortest published SNP map in apple, while including the largest number of markers, families and individuals. This result validates our methodology, proving its value for the construction of integrated linkage maps for any outbreeding species.

No MeSH data available.