Statistical Models for Detecting Gene Conversion

研究者开发用于检测基因转换的统计测试。基因转换是生物学中导致DNA片段被复制到另一个DNA片段的任何过程。基因转换是进化过程中的一个重要过程，也对人类健康产生影响。例如，它可以导致致病性性状在细菌之间转移，病毒基因组的部分在病毒之间转移。统计测试检测基因转换在对齐的DNA或氨基酸序列进行了研究和扩展。特别强调排列或类似排列的测试，这些测试可以评估两个序列相似的异常长片段，或者单个序列异常发散的片段。（这包括Karlin-Altschul类型的测试，这种测试执行起来更快，但不如排列测试准确。）该建议的一些具体目标是：（一）扩大这些测试，通过寻找相似分数而不是片段长度的高分片段来允许偶尔的不匹配和比对差距；（二）制定方法来提高这些测试的效力和有效性；（三）比较检测基因转换的统计程序的效率和效力；（四）在一个网站上向生物界提供改进后的软件。人类艾滋病毒和流感病毒属于病毒家族（RNA逆转录病毒），不仅高度可变，而且它们之间也可以交换遗传物质片段。能够交换遗传物质块的病毒更危险，因为病毒中适应性更强的部分可以迅速传播，而且只针对部分病毒的疫苗可能会迅速过时。在这个病毒家族中也有几个植物病毒科，它们对经济上重要的粮食作物造成损害。该项目扩展了一种统计测试，可以从基因序列样本中检测出这种交换的历史，并为最可能发生的事件分配统计可能性度量。这将有助于了解这些特定病毒进化的本质。这项工作扩展了研究者早期的工作，在假设交换的物质没有经历后来的突变的情况下，在对齐的DNA序列中寻找可能的交换事件。这种假设并不适用于这些高度易变的病毒。此外，最近的研究表明，早期的测试在某些程序中存在偏见，可能会隐藏一些交换（或基因转换）事件的证据。这种偏见被消除了。最后，执行改进程序的计算机软件，包括列出最有可能的交换事件及其统计可能性，在互联网上为生物社区提供了更多的可用性。（实现上述步骤的软件已被广泛使用。）

Sawyer 9707045 The investigator develops statistical tests for detecting gene conversion. Gene conversion is any process in biology that causes a segment of DNA to be copied to another DNA segment. Gene conversion is an important process in evolution, and also has consequences for human health. For example, it can cause the transference of pathogenic traits between bacteria and parts of viral genomes between viruses. Statistical tests for detecting gene conversion in aligned DNA or amino acid sequences are studied and extended. Particular emphasis is given to permutation or permutation-like tests that can evaluate unusually long segments on which two sequences are similar, or else on which a single sequence is unusually divergent. (This includes tests of Karlin-Altschul type, which are faster to perform but are not as accurate as permutation tests.) Some specific aims of the proposal are: (i) to extend these tests to allow occasional mismatches and alignment gaps by finding high-scoring segments for a similarity score rather than segment length, (ii) develop ways to improve the power and effectiveness of these tests, (iii) compare statistical procedures for detecting gene conversion in regards to efficiency and power, and (iv) make the resulting improved software available to the biological community on a Web site. Human HIV and influenza viruses belong to a family of viruses (RNA retroviruses) that not only are highly changeable, but can also exchange pieces of genetic material among themselves. Viruses that can exchange blocks of genetic material are more dangerous since the more adaptive parts of the virus can spread rapidly, and since vaccines that target only part of the virus can become rapidly obsolete. There are also several families of plant viruses in this viral family that cause damage to food crops of economic importance. The project extends a statistical test that can detect a history of such exchanges from a sample of genetic sequ ences and assign measures of statistical likelihood to the most likely events. This will help to understand the nature of the evolution of these particular viruses. The work extends earlier work of the investigator that looked for possible exchange events in aligned DNA sequences under the assumption that the exchanged material had not undergone later mutation. This assumption does not hold for these highly mutable viruses. In addition, recent work has shown that the earlier test has biases in some of its procedures that can hide evidence for some exchange (or gene conversion) events. This bias is removed. Finally, computer software that implements the improved procedures, including listing the most likely exchange events and their statistical likelihood, is made more available on the internet for the biological community. (Software that implements the earlier procedures has been widely used.)