不同统计遗传模型QTL定位方法应用效果的模拟比较

分子遗传和数量遗传的结合,发展了QTL定位研究。随着定位方法与软件的建立和完善,QTL定位的研究越来越多。准确定位的QTL可用于分子标记辅助选择和图位克隆,而假阳性QTL将误导定位信息的应用。本文分析了迄今主要定位方法(软件)对于各种遗传模型数据的适用性。应用计算机模拟4类遗传模型不同的重组自交系群体(RIL),第一类只包含加性QTL;第二类包含加性和上位性互作QTL;第三类包含加性QTL和QTL与环境互作效应;第四类包含加性、上位性互作QTL和QTL与环境互作效应。每类按模拟QTL个数不同设两种情况,共分为8种数据模型(下称M-1～M-8)。选用WinQTLCart 2.5的复合区间作图(下称CIM)、多区间作图前进搜索(MIMF)、多区间作图回归前进选择(MIMR)、IciMapping 2.0的完备复合区间作图(ICIM)、MapQTL 5.0的多QTL模型(MQM)以及QTLnetwork 2.0的区间作图(MCIM)6种程序对8种不同遗传模型的RIL进行QTL检测。结果表明,不同程序适用的遗传模型范围不同。CIM和MQM只适于检测第一类模型;MIMR、MIMF和ICIM只适于检测第一类和第二类模型;只有MCIM适于检测所有4类遗传模型;因而不同遗传模型数据的最适合检测程序不同。由于未知实际数据的遗传模型,应采用在复杂模型程序,如QTLnetwork 2.0,扫描基础上的多模型QTL定位策略,对所获模型用相应模型软件进行验证。

Simulation Comparisons of Effectiveness among QTL Mapping Procedures of Different Statistical Genetic Models

QTL mapping has emerged based on the development and integration of molecular genetics and quantitative genetics. Along with the establishment and improvement of QTL mapping procedures, a great number of studies of QTL mapping in various crop species have been carried out. QTLs detected with high accuracy can be used for marker-assisted selection and map-based cloning, while the false-positive QTLs are meaningless, even mislead their usefulness. In the present study, the recombinant inbred line (RIL) populations were simulated based on four kinds of genetic models, including Model I, additive QTL; Model II, additive and epistatic QTLs; Model III, additive QTL and QTL×Environment interaction, and Model IV, additive, epistatic QTLs and QTL×Environment interaction. Two sets of RIL data for each of the four models were obtained, in a total of eight sets of RIL data designated as M-1~M-8. Six QTL mapping procedures, i.e. CIM (Composite interval mapping), MIMF (forward search of multiple interval mapping) and MIMR (regression forward selection of multiple interval mapping) of WinQTL Cartographer Version 2.5, ICIM (Inclusive composite interval mapping) of IciMapping Version 2.0, MQM (multiple-QTL model) of MapQTL Version 5.0, and NWIM (interval mapping) of QTLnetwork Version 2.0 were used for detecting QTLs of the eight sets of RIL data. The results showed: (1) Different mapping procedures fit different genetic models. CIM and MQM were only suitable for Model I data. MIMR, MIMF and ICIM were only suitable for Model I and Model II data. Only NWIM was suitable for all four models’ data. Therefore, the data from different genetic models corresponded to different optimal QTL mapping procedures. (2) Since the genetic model of the practical experimental data was unknown, a multiple model mapping strategy should be taken, i.e. a full model scanning with complex model procedure, such as QTLnetwork 2.0, followed by verification with another procedure corresponding to the scanning results.