New ideas for plant breeding Make effective use of genetic diversity in seed bank. Nature Plant paper recommended.

The new study, using sorghum as an example, shows breeders how to effectively tap and utilize genetic diversity from plant gene pools.
Writer Patrick J. Brown compiles a well-designed tunnel system deep underground on a Small Norwegian island, the Svalbard Global Seed Vault, which currently holds more than 800,000 plant samples and has a capacity of 4.5 million, as a "backup" of the genetic material needed to sustain human survival.
Seed Bank Entrance The plant gene pool worldwide holds millions of crop varieties, some of which can undoubtedly be used to address future droughts, pests and diseases and the spread of crop diseases, but most of these species are buried in the reservoir and not used by breeding projects.
can we help breeders acquire these genetic diversities that are very useful to them? As more gene pools add information about genotypes, this may be achieved through genomic prediction.
a recent study published in Nature Plants, yu and others, for the first time demonstrated that genetic profiles in the sorghum gene pool can be accurately predicted through genome prediction.
predictions are based on the genotypes of known individuals and infer their esotype data.
inference process is based on a model built from a "training set" of individuals that contain esoterype and genotype information.
specifically, the model assigned esoteric effects to each allegable gene in a tagged data set, integrated local management and environment, and then calculated and calculated the esotericity of each individual.
Yu and his colleagues genotype 1,000 light-cycle sensitive tropical sorghum species in the USDA gene pool and used the data to select a representative training set of nearly 300 varieties to measure biological yields and their associated features, including plant height and resilience to inverting.
data set was then used to predict which of the remaining nearly 700 varieties produced the lowest and which had the highest yields.
this predictive model validates 200 varieties with real esodype data, including 50 varieties with the highest and lowest forecast yields and 100 randomly selected varieties.
accuracy of the prediction of biological yield (observation and prediction esothroid correlation, r) is 0.76, which is very high for complexity such as crop yield.
encouraging results provide the basis for a series of future studies.
Yu's study, higher predictive accuracy may be due to their use of highly diverse species resources.
problem with genome prediction is the influence of group structures, networks that hide common ancestors between different individuals.
population structures are tightly controlled in genome-wide association studies (GWAS), but genomic predictions are not.
sorghum is divided into five small species (race, subsethagical subseeds of classification units) (bicolor, Caudatum, durra, guinea, and kafir, in yu et al., these different small species were highly correlated with biological yields, with durras and bicolors having the largest and smallest yields, respectively, and the accuracy of their prediction models came in part from simply predicting whether these varieties belonged to durras or bicolors, while trained breeders could make similar predictions without genotype data.
their prediction models may overemphamphaly emphasize subseed differences, but for breeders, the need to maintain crop diversity and long-term genetic reserves requires analysis of all small species rather than focusing only on the highest yields of durras.
limitation of genome prediction is the keeping of genetic diversity, Yu et al. wrote in a second validation experiment using a completely different set of 580 sorghum lineits.
the accuracy of predicting biological yields in this independent concentration decreased significantly (r s 0.56), possibly because this new data set contains special allethic genes that are not found in many training sets.
, although the training set was selected to represent the 1,000 symbols of the original set, it was difficult to represent the newly created stand-alone set, resulting in a decrease in predictive accuracy.
genetic diversity of most species is due to the hiding of rare alletic genes, which are often overlooked in a given training set because of their low frequency of appearance.
solution to this problem includes weighting the effects of these allegens and selecting a subset of the genome rather than the entire genome.
this strategy has the potential to uncover surprising and valuable allebean genes, but the allebeans missing from training sessions are still not recognizable.
results show that the design of training sets has a great influence on genome prediction results. the validation of
genomic prediction models can be self-reinforcing, and the 50 varieties used to validate esoteric patterns in the highest and lowest predicted yields are usually genotypes that are very similar to training sets because these individuals have a higher degree of reliability than those with special esoterics.
to deal with this bias, Yu et al. applied a new measure, the reliability cap, to prioritize the inclusion of new, genetically specific individuals in the training set.
the development of predictive models is an iterative process in which each generation of esothype data shows the shortcomings of the model (poorly predicted individuals) so that further improvements can be made.
in this first iteration, Yu et al. found that the highest biological yield measured in the validation set was not in the 50 most predicted series, but in those 100 randomly selected series.
Yu and others have shown that we can begin to make genome predictions, more needs to be done to ensure that the great potential of the gene pool is used most effectively.
author Patrick J. Brown is an associate professor in the Department of Crop Science at the University of Illinois. McCouch, S. et al. Nature 499, 23-24 (2013).2. Yu, X. et al. Nat. Plants 2, 16150 (2016).3. Jannink, J.-L. Genet. Sel. Evol. 42, 35(2010).4. Daetwyler, H. D., Hayden, M. J., Spangenberg, G. C. and Hayes, B. J. Genetics 200, 1341-1348 (2015).5. Karaman, E., Cheng, H., Firat, M. Z., Garrick, D. J. and Fernando, R. L. PLoS ONE 11, e0161054 (2016).Related links: Related paper information Title Genomic prediction contributing to agaga global strategy to turbocharge gene banks Authors Xiaoqing Yu, Xianran Li, Tingting Guo, Chengsong Zhu, Yuye Wu, Sharon E. Mitchell, Kraig L. Roozeboom, Donghai Wang, Ming Li Wang, Gary A. Pederson, Tesfaye T. T. Tesso, Patrick S. Schnable, Rex Bernardo and Jianming Yu Journal Nature Plants Digital Identification Code doi:10.1038/nplants.2016.150 Published online:03 October 2016 Summary The 7.4 million plant accessions in gene banks are largely underutilized due to various resources constraints, but current genomic and analytics are enabling us to mine this natural heritage. Here we report a proof-of-concept study to integrate genomic prediction into a broad germplasm evaluation process. First, a set of 962 biomass sorghum accessions were chosen as a reference set by germplasm curators. With high-high-thiontyping-by-sequencing (GBS), we genetically provided this reference set with 340,496 single nucleotide polyphisms (SNPs). A set of 299 accessions was selected as the training set to represent the overall diversity of of the set reference, and we phenotypically directed the training set for biomass yield and other related traits. Cross-validation with multiple analytical methods using the data of this training set development high predictioncy for biomass yield. Empirical competitions with a 200-accession validation set chosen from the reference set confirmed high prediction accuracy. The potential to apply the prediction model to broader genetic contexts was also examined with an independent population. Detailed analyses on predictionliability re provided new insights into strategy optimization. The success of this project designs that a global, cost-effective strategy may be designed to assess the vast amount of value germplasm archived in 1,750 gene banks. Yu, et al. Nat. Plants 2, 16150 (2016) Source: Science Circle.