【作者】 袁雄；

【导师】 袁哲明；

【作者基本信息】 湖南农业大学 ， 生物信息学， 2020， 博士

【摘要】 单核苷酸多态性(SNP)是种质资源基因组中发现的最丰富的序列变异,已广泛用作分子遗传连锁图谱中的多样性分析和品种鉴定。随着分子标记辅助选择育种、基因组育种等快速育种技术的发展,品种数量快速增加,现有选择核心标记区分品种的方法难以满足所有品种真实性鉴定的需求。为保障育种者知识产权、加强种质管理、促进新品种的引入与遗传改良,亟需发展普适性的基于SNP标记的种质资源鉴定方法。本研究首先以公开的水稻3k基因组计划数据集为研究对象(训练集),根据公开的已克隆水稻基因,筛选位于基因编码区的SNP位点,采用新发展的条件随机法、基于多态性信息含量(PIC)的决策树法和基于贪婪策略的复杂度法高效地筛选出可鉴定全部参试材料的SNP组合,并将多套精简的SNP标记组合求并集,同时构建多核苷酸多态性(MNP)标记组合,对品种进行联合鉴定,以提高鉴定方法的容错能力;再以公开的4591(含1570个否定品种,即与训练集中的品种不同)和自测的343(均为肯定品种,即与训练集中的品种相同)两个水稻种质数据集为测试集,进行不同方法产生的不同数量核心SNP标记组合的品种鉴定效能研究;最后以公开的302份大豆品种为研究对象,进行条件随机法和决策树法筛选核心标记鉴定品种的普适性研究。主要研究结果如下:(1)利用R语言平台,新开发了基于“随机”、“贪婪”、“分而治之”组合策略的条件随机法、基于高PIC的决策树法和基于贪婪策略的复杂度法等三种基于全基因组高效筛选核心SNP标记鉴定种质资源的方法,三种方法区分3021个品种最精简的SNP数量分别为63、69和74个。将多套精简的SNP标记组合构成并集和构建MNP标记来对品种进行联合鉴定,能有效提高鉴定方法的容错能力,改变15%的待测品种标记基因型,条件随机法335个核心SNP标记组合、决策树法292个核心SNP标记组合、复杂度法311个核心SNP标记组合、条件随机法与决策树法两种方法455个核心标记的并集组合以及153个核心标记的MNP标记组合仍分别有97.4%、98.2%、98.4%、98.2%、98.7%的准确度。(2)各核心SNP标记组合的品种区分能力随着核心标记数量的增加而上升,但呈现先快速上升后逐步趋于稳定的趋势,当核心标记数量达到335个时,其品种鉴定能力趋于稳定,进一步将核心标记数量增加到455个和601个,其品种区分能力没有进一步上升,仍然是区分4591个测试集品种中的4587个。若采取高PIC法区分测试集中水稻品种,即使用所有的高PIC值SNP标记组成并集,也只能区分其中的4433个水稻品种,其鉴定能力都不如新方法筛选的最精简的核心SNP标记组合,说明新方法筛选出的多组核心标记并集,对否定品种的鉴定效率高。(3)将343个来源于3K水稻基因组种质的品种种植于武汉地区,以其重测序的SNP基因型数据作为独立测试集,结果在遗传相似性大于95%的品种中,条件随机法335个核心SNP标记组合、决策树法292个核心SNP标记组合、复杂度法311个核心SNP标记组合、条件随机法与决策树法455个核心标记的并集组合以及601个核心标记的MNP标记组合,独立预测准确率分别为87.00%、87.50%、87.36%、88.24%、88.06%。(4)条件随机法和决策树法筛选的核心SNP标记并集组合兼具两种方法筛选的核心SNP标记组合的优点,用其455个SNP标记构建水稻指纹图谱,可用于水稻种质资源的高效鉴定。(5)用条件随机法和决策树法分别筛选13个SNP标记构建核心SNP组合,在302份大豆种质中均能形成302个特异单倍型,成功鉴定所有供试材料。本研究开发的三种核心SNP标记筛选方法,能高效筛选出较少的SNP标记鉴定全部供试材料,节约鉴定成本,与其他现有方法相比,对新品种材料具有更强的区分能力。三种方法具有很强的推广性和可移植性,可广泛应用于多个物种的品种鉴定,为动植物品种鉴定提供了新思路。更多还原

【Abstract】 Single nucleotide polymorphism(SNP)is the most abundant sequence variation found in the genome of germplasm resources,which has been widely used for diversity analysis and variety identification in molecular genetic linkage map.With the rapid development of molecular marker assisted selection and genome breeding,the number of varieties is increasing rapidly.The existing methods of selecting core markers to distinguish varieties are difficult to meet the needs of authenticity identification of all varieties.In order to protect the intellectual property rights of breeders,strengthen germplasm management,and promote the introduction and genetic improvement of new varieties,it is urgent to develop a universal identification method based on SNP markers.In this study,firstly,the open genotype data set of rice 3K genome project was selected as training set.According to the published cloned rice genes,SNPs located in the gene coding region were screened.Three novel methods,conditional random(CR)method,the decision tree method based on polymorphism information content(PIC)(PIC-DT)and the complexity method based on greedy strategy(GS-C)were developed to screen the SNP combinations for all the tested materials identification.In order to reduce identification error ratio,several core SNP marker combination sets were united and multi nucleotide polymorphism(MNP)markers combinations were constructed for varieties identification.Then,4591(including 1570 uncertain varieties,which are different from the varieties in the training set)and 343(all are definite varieties,which are from the varieties in the training set)self-resequencing rice germplasm data sets were used to perform variety identification efficiency research on different numbers of core SNP marker combinations produced by different methods.Finally,with 302 public soybean varieties SNP genotype data set as the research object,the universality research of the conditional random method to screen core markers to identify varieties was carried out.The main results were as follows:Based on the R language platform,we developed three methods to identify germplasms based on genome-wide efficient screening core SNP markers,including conditional random method based on "random","greedy" and "divide and conquer" strategy,decision tree method based on high PIC and complexity method based on greedy strategy.The three methods can separately distinguish 3021 varieties with 63,69 and 74 SNPs.Combining multiple sets of SNP marker combinations and constructing MNP markers to jointly identify varieties could effectively improve thefault tolerance ability of identification methods.When changing 15% of the genotypes of tested varieties,the accuracy of 335 core SNP marker combinations in conditional random method,292 core SNP marker combinations in decision tree method,311 core SNP marker combinations in complexity method,455 core markers by uniting conditional random method and decision tree method,and 153 core MNP markers were 97.4%,98.2%,98.4%,98.2% and 98.7%,respectively.The variety distinguishing ability of each core SNP marker combination increases with the increase in the number of core markers,but showed a trend of rapid increase first and then gradually stabilized.When the number of core SNP markers reached 335,the ability of variety identification tended to be stable.When the number of core markers was increased to 455 and 601,the ability of distinguishing varieties did not rise further,and it was still identification 4587 varieties in 4591 data set.All the SNP markers with high PIC value only 4433 rice varieties were distinguished.The identification ability of high PIC method was not as good as new developed methods.It indicated that the combination of multiple core markers screened by the three new method has high identification efficiency for uncertain varieties.Three hundred and forty-three rice varieties derived from 3K rice genome germplasm were planted in Wuhan.The SNP genotypes dataset were used as independent test set.The results showed that among the varieties with genetic similarity greater than 95%,the accuracy of 335 core SNP marker combinations selected with conditional random method,292 core SNP marker combinations selected with decision tree method,311 core SNP marker combinations selected with complexity method,455 core SNP selected with conditional random method and decision tree method,and 292 MNP markers(contains 601 SNPs),were 87.00%,87.50%,87.36%,88.24% and 88.06%,respectively.The combination of core SNP markers screened by conditional random method and decision tree method has the advantages of the core SNP marker combinations screened by the two methods.Using its 455 SNP markers to construct a rice fingerprint map was used for efficient identification of rice germplasm resources.By using CR and PIC-DT methods,two combinations each containing 13 SNP markers were separately selected and successfully identified 302 soybean germplasms.The three core SNP marker screening methods developed in this research efficiently screened out fewer SNP markers for all test materials identification,which save identification costs,and have a stronger ability to distinguish new varieties of materials than other existing methods.The three methods have strong generalization and portability,can be widely used in the varieties/cultivars identification of multiple species,and provide a new theoretical for animal and plant varieties identification.更多还原