BackgroundEnvironmental adaptation and expanding harvest seasons are primary goals of most peach [Prunus persica (L.) Batsch] breeding programs. Breeding perennial crops is a challenging task due to their long breeding cycles and large tree size. Pedigree-based analysis using pedigreed families followed by haplotype construction creates a platform for QTL and marker identification, validation, and the use of marker-assisted selection in breeding programs.ResultsPhenotypic data of seven F1 low to medium chill full-sib families were collected over two years at two locations and genotyped using the 9K SNP Illumina array. Three QTLs were discovered for bloom date (BD) and mapped on linkage group 1 (LG1) (172 – 182 cM), LG4 (48 – 54 cM), and LG7 (62 – 70 cM), explaining 17-54%, 11-55%, and 11-18% of the phenotypic variance, respectively. The QTL for ripening date (RD) and fruit development period (FDP) on LG4 was co-localized at the central part of LG4 (40 - 46 cM) and explained between 40-75% of the phenotypic variance. Haplotype analyses revealed SNP haplotypes and predictive SNP marker(s) associated with desired QTL alleles and the presence of multiple functional alleles with different effects for a single locus for RD and FDP.ConclusionsA multiple pedigree-linked families approach validated major QTLs for the three key phenological traits which were reported in previous studies across diverse materials, geographical distributions, and QTL mapping methods. Haplotype characterization of these genomic regions differentiates this study from the previous QTL studies. Our results will provide the peach breeder with the haplotypes for three BD QTLs and one RD/FDP QTL for the creation of predictive DNA-based molecular marker tests to select parents and/or seedlings that have desired QTL alleles and cull unwanted genotypes in early seedling stages.

26High-quality genotypic data is a requirement for many genetic analyses. For any crop, errors in genotype 27 calls, phasing of markers, linkage maps, pedigree records, and unnoticed variation in ploidy levels can 28 lead to spurious marker-locus-trait associations and incorrect origin assignment of alleles to individuals. 29High-throughput genotyping requires automated scoring, as manual inspection of thousands of scored 30 loci is too time-consuming. However, automated SNP scoring can result in errors that should be 31 corrected to ensure recorded genotypic data are accurate and thereby ensure confidence in 32 downstream genetic analyses. To enable quick identification of errors in a large genotypic data set, we 33 have developed a comprehensive workflow. This multiple-step workflow is based on inheritance 34 principles and on removal of markers and individuals that do not follow these principles, as 35 demonstrated here for apple, peach, and sweet cherry. Genotypic data was obtained on pedigreed 36 germplasm using 6-9K SNP arrays for each crop and a subset of well-performing SNPs was created using 37 ASSIsT. Use of correct (and corrected) pedigree records readily identified violations of simple inheritance 38 principles in the genotypic data, streamlined with FlexQTL TM software. Retained SNPs were grouped into 39 haploblocks to increase the information content of single alleles and reduce computational power 40 needed in downstream genetic analyses. Haploblock borders were defined by recombination locations 41 detected in ancestral generations of cultivars and selections. Another round of inheritance-checking was 42 conducted, for haploblock alleles (i.e., haplotypes). High-quality genotypic data sets were created using 43 this workflow for pedigreed collections representing the U.S. breeding germplasm of apple, peach, and 44 sweet cherry evaluated within the RosBREED project. These data sets contain 3855, 4005, and 1617 45 SNPs spread over 932, 103, and 196 haploblocks in apple, peach, and sweet cherry, respectively. The 46 highly curated phased SNP and haplotype data sets, as well as the raw iScan data, of germplasm in the 47 apple, peach, and sweet cherry Crop Reference Sets is available through the Genome Database for 48 Rosaceae. 3 49 50 103 Manual p17 [34]). The presence of one or more additional SNPs, insertions, or deletions in the probe-104 binding region can lead to reduced or loss of binding affinity for the SNP's probe and thereby to the 105 presence of additional clusters, both of which can lead to incorrect genotype scoring of some SNPs [33]. 106 107 No systematic workflow exists to efficiently detect and resolve all types of errors from a genotypic data 108 set for pedigreed germplasm. Methods and software exist to tackle specific types of errors. For example, 109 the ASSIsT software was developed for use with Illumina Infinium® arrays to identify which SNPs show 110 robust results, which SNPs might have genotype calling errors due to alleles with reduced affinity or null 111 alleles, and which...

Linkage mapping is an approach to order markers based on recombination events. Mapping algorithms cannot easily handle genotyping errors, which are common in high-throughput genotyping data. To solve this issue, strategies have been developed, aimed mostly at identifying and eliminating these errors. One such strategy is SMOOTH, an iterative algorithm to detect genotyping errors. Unlike other approaches, SMOOTH can also be used to impute the most probable alternative genotypes, but its application is limited to diploid species and to markers heterozygous in only one of the parents. In this study we adapted SMOOTH to expand its use to any marker type and to autopolyploids with the use of identity-by-descent probabilities, naming the updated algorithm Smooth Descent (SD). We applied SD to real and simulated data, showing that in the presence of genotyping errors this method produces better genetic maps in terms of marker order and map length. SD is particularly useful for error rates between 5% and 20% and when error rates are not homogeneous among markers or individuals. With a starting error rate of 10%, SD reduced it to ∼5% in diploids, ∼7% in tetraploids and ∼8.5% in hexaploids. Conversely, the correlation between true and estimated genetic maps increased by 0.03 in tetraploids and by 0.2 in hexaploids, while worsening slightly in diploids (∼0.0011). We also show that the combination of genotype curation and map re-estimation allowed us to obtain better genetic maps while correcting wrong genotypes. We have implemented this algorithm in the R package Smooth Descent.

Unordered parent-offspring (PO) relationships are an outstanding issue in pedigree reconstruction studies. Resolution of the order of these relationships would expand the results, conclusions, and usefulness of such studies; however, no such PO order resolution (POR) tests currently exist. This study describes two such tests, demonstrated using SNP array data in the outcrossing species apple (Malus × domestica) on a PO relationship of known order (“Keepsake” as a parent of “Honeycrisp”) and two PO relationships previously ordered only via provenance information. The first test, POR-1, tests whether some of the extended haplotypes deduced from homozygous SNP calls from one individual in an unordered PO duo are composed of recombinant haplotypes from accurately phased SNP genotypes from the second individual. If so, the first individual would be the offspring of the second individual, otherwise the opposite relationship would be present. The second test, POR-2, does not require phased SNP genotypes and uses similar logic as the POR-1 test, albeit in a different approach. The POR-1 and POR-2 tests determined the correct relationship between “Keepsake” and “Honeycrisp”. The POR-2 test confirmed “Reinette Franche” as a parent of “Nonpareil” and “Brabant Bellefleur” as a parent of “Court Pendu Plat”. The latter finding conflicted with the recorded provenance information, demonstrating the need for these tests. The successful demonstration of these tests suggests they can add insights to future pedigree reconstruction studies, though caveats, like extreme inbreeding or selfing, would need to be considered where relevant.

BackgroundSingle nucleotide polymorphism (SNP) array technology has been increasingly used to generate large quantities of SNP data for use in genetic studies. As new arrays are developed to take advantage of new technology and of improved probe design using new genome sequence and panel data, a need to integrate data from different arrays and array platforms has arisen. This study was undertaken in view of our need for an integrated high-quality dataset of Illumina Infinium® 20K and Affymetrix Axiom® 480K SNP array data in apple (Malus × domestica). In this study, we qualify and quantify the compatibility of SNP calling, defined as SNP calls that are both accurate and concordant, across both arrays by two approaches. First, the concordance of SNP calls was evaluated using a set of 417 duplicate individuals genotyped on both arrays starting from a set of 10,295 robust SNPs on the Infinium array. Next, the accuracy of the SNP calls was evaluated on additional germplasm (n=3,141) from both arrays using Mendelian inconsistent and consistent errors across thousands of pedigree links. While performing this work, we took the opportunity to evaluate reasons for probe failure and observed discordant SNP calls.ResultsConcordance among the duplicate individuals was on average of 97.1% across 10,295 SNPs. Of these SNPs, 35% had discordant call(s) that were further curated, leading to a final set of 8,412 (81.7%) SNPs that were deemed compatible. Compatibility was highly influenced by the presence of alternate probe binding locations and secondary polymorphisms. The impact of the latter was highly influenced by their number and proximity to the 3’ end of the probe.ConclusionsThe Infinium and Axiom SNP array data were mostly compatible. However, data integration required intense data filtering and curation. This work resulted in a workflow and information that may be of use in other data integration efforts. Such an in-depth analysis of array concordance and accuracy as ours has not been previously described in literature and will be useful in future work on SNP array data integration and interpretation, and in probe/platform development.

Linkage mapping is an approach to order markers based on recombination events. Mapping algorithms cannot easily handle genotyping errors, which are common in high-throughput genotyping data. To solve this issue, strategies have been developed, aimed mostly at identifying and eliminating these errors. One such strategy is SMOOTH (van Os et al. 2005), an iterative algorithm to detect genotyping errors. Unlike other approaches, SMOOTH can also be used to impute the most probable alternative genotypes, but its application is limited to diploid species and to markers heterozygous in only one of the parents. In this study we adapted SMOOTH to expand its use to any marker type and to autopolyploids with the use of identity-by-descent probabilities, naming the updated algorithm Smooth Descent (SD). We applied SD to real and simulated data, showing that in the presence of genotyping errors this method produces better genetic maps in terms of marker order and map length. SD is particularly useful for error rates between 5% and 20% and when error rates are not homogeneous among markers or individuals. Moreover, the simplicity of the algorithm allows thousands of markers to be efficiently processed, thus being particularly useful for error detection in high-density datasets. We have implemented this algorithm in the R package SmoothDescent.

Quantitative trait locus (QTL) analysis allows to identify regions responsible for a trait and to associate alleles with their effect on phenotypes. When using biallelic markers to find these QTL regions, two alleles per QTL are modelled. This assumption might be close to reality in specific biparental crosses but is unrealistic in situations where broader genetic diversity is studied. Diversity panels used in genome-wide association studies or multi-parental populations can easily harbour multiple QTL alleles at each locus, more so in the case of polyploids that carry more than two alleles per individual. In such situations a multiallelic model would be closer to reality, allowing for different genetic effects for each potential allele in the population. To obtain such multiallelic markers we propose the usage of haplotypes, concatenations of nearby SNPs. We developed “mpQTL” an R package that can perform a QTL analysis at any ploidy level under biallelic and multiallelic models, depending on the marker type given. We tested the effect of genetic diversity on the power and accuracy difference between bi-allelic and multiallelic models using a set of simulated multiparental autotetraploid, outbreeding populations. Multiallelic models had higher detection power and were more precise than biallelic, SNP-based models, particularly when genetic diversity was higher. This confirms that moving to multi-allelic QTL models can lead to improved detection and characterization of QTLs. Key message QTL detection in populations with more than two functional QTL alleles (which is likely in multiparental and/or polyploid populations) is more powerful when using multiallelic models, rather than biallelic models.

No abstract