用SnaP Shot的方法对线粒体7146位点进行基因分型

利用SNaP SHot的方法对线粒体7146位点在7300人的样品中进行基因分型，这个位点在非洲人群中存在2种等位基因，但是在非洲人群中仅存在突变型，这个研究对利用线粒体研究人类进化，特别是对东亚人群的起源有较大的应用价值，同时还建立了一套大规模，高通量的SNP分型技术平台。     

Molecular mechanism of the inhibitory effect of methylation outside the recognition sequence of restriction endonuclease Pvu II on its cleavage activity

In 1991, we found that methylation outside the Pvu II recognition sequence could partially inhibit its cleavage activity. To clarify the molecular mechanism, three plasmids with different methylation states were constructed. Then, together with the original one, four plasmids were digested with different amounts of Pvu II. Results show that methylation on both sites results in 90 % inhibition; moving the methylated site one base further away decreases the inhibitory effect to about 30 %; with the adjacent dam methylation site eliminated, the inhibitory effect disappears. The data suggest that the inhibition of cleavage activity caused by outside methylation is not “all or none”, and the degree of inhibition is dependent on the position and the number of methylated bases.     

基于黄金分割律选择算子的改进进化编程算法及其在机器人路径规划中的应用

通过分析及结合机器人路径规划的进化编程仿真实验指出,保存最优个体或淘汰最差个体都会引起进化算法早熟现象,并提出如下改进算法按适应度从高(优)到低,依黄金分割律分别取出优秀和最差部分个体(染色体),通过赌轮法,并按适者生存原则从优秀的部分中选择一个体直接进入子代,而按劣者淘汰原则从最差部分中选择一个体从子代中排除.将改进算法运用于机器人路径规划,通过仿真实验检验了算法的有效性.     

分子序列的突变危险性与遗传密码的编码规则

定义了分子序列（碱基序更和对应的氨基酸序列）的突变危险性函数，通过危险性函数的极小化可以导出遗传密码的简并规则，终止密码子在密码表上的位置及亲－水畴排布，氨基酸在密码表上的定位接近标准密码表，各个生物类别的突变危险性函数值差别不超过５％，可看作守恒量，Ｒａｓ家庭癌基因产物具有较大的突变危险性函数值，这提示分子序列的突变危险性函数可能有重要的应用价值和理论意义。     

Study on MCM Interconnect Test Generation Based on Ant Algorithm with Mutation Operator

Introduction With rapid development of very large scale in-tegration(VLSI),multi-chip module(MCM)andmulti-layer printed circuit boards(MPCB),inter-connect test technology has become a bottleneckinthe application of these circuits.The high reliabili-ty of MCMis due to that bare integrated circuitchips are welded and interconnected under highdensity and small di mension conditions[1].Testgenerationis one key technologies of MCMinter-connect test,so study on novel method of test gen-eratio…     

遗传算法在数据内部排序上的应用

用遗传算法进行数据内部排序，可以打破传统排序算法在时间复杂度方面所存在的下限，因为求适值部分可并行处理，对包括适值函数、杂交、变异等在内的一些关键问题进行了研究，尤其就不同适值函数的设计对算法性能等方面的影响进行了对比讨论。     

Applying genetic algorithms to the study of protein mutation theory

Genetic algorithms were applied to the study of simulation of protein mutation carried out on two_dimensional lattice model; to the study of effects of single mutation and double mutation on protein folding and protein structure stability. It is found that in two_dimensional lattice models, replacement of inner core hydrophobic residue by hydrophilic residue will result in reduction of protein stability; at the same time the number of residues of protein surface relatively grows with increase of protein chain length; and most mutations occur in residues of protein surface, these mutations are all neutral and have no effects on protein natural structure. The two mutations of double mutation are interactive and related to each other.     

遗传算法在前馈神经网络中的应用

比较了遗传算法和神经网络的特点，对于将遗传算法用于前馈神经网络的可能性进行了研究，并给出了应用实例；同时提出了一种构造变长染色体用于神经网络演化的方法，并描述了作用于变长染色体上的遗传操作机制     

基于自适应变异因子策略的混合蛙跳算法

由于基本混合蛙跳算法在对问题的优化求解中存在着收敛速度慢、优化精度低且容易陷入局部最优等问题,因此提出了一种新的混合蛙跳算法。对基本混合蛙跳算法的组内更新策略进行重新设计,引入自适应变异因子来控制青蛙的移动步长;在算法中将改进的粒子群优化算法有机地嵌入其中,这样算法在搜索过程中就增加了发现新解的概率,维持了种群的多样性,从而使算法不易陷入局部最优。通过对标准函数进行优化测试,结果证明其具有良好的优化性能。     

改进的免疫算法及其在函数优化中的应用

为了提高免疫算法的搜索能力,根据生物免疫机制及生物进化的周期性,设计了一种周期变化变异算子。为了避免仅仅以亲和度作为免疫选择评价标准,低亲和度抗体过度抑制,提出了将抗体浓度引入到亲和度中作为评价指标,设计了一种改进的免疫选择算子。基于马尔科夫链,分析了改进免疫算法的收敛性。为了测试该算法的有效性,将算法应用于函数优化问题中。仿真结果表明,改进的免疫算法具有更高的搜索速度和精度。