Rzhetsky A, Morozov P
Columbia Genome Center, Columbia University, 1050 St. Nicholas Avenue, Unit 109, New York, NY 10032, USA.
Pac Symp Biocomput. 2001:203-14. doi: 10.1142/9789814447362_0021.
We suggest a method implemented in a computer program, immodestly dubbed TSUNAMI, that allows us to compare two homologous protein subfamilies with respect to the distribution of substitution rates along sequences. This study furthers our earlier work on a wavelet model of rate variation (1). The current approach allows sensitive detection of subtle discordances in the selection patterns between two protein subfamilies. In addition to performing fast computation of the maximum posterior probability estimates of the relative substitution rates, the method can select the most appropriate number of wavelet parameters for a particular dataset. TSUNAMI is based on a Markov chain Monte Carlo technique, and appears to be more applicable to larger datasets than is the full likelihood-based approach.
我们提出了一种在计算机程序中实现的方法，该程序被毫不谦虚地命名为海啸（TSUNAMI），它使我们能够就沿序列的替换率分布比较两个同源蛋白质亚家族。这项研究推进了我们早期关于速率变化的小波模型的工作(1)。当前的方法能够灵敏地检测两个蛋白质亚家族之间选择模式中的细微不一致。除了能快速计算相对替换率的最大后验概率估计值外，该方法还能为特定数据集选择最合适数量的小波参数。海啸（TSUNAMI）基于马尔可夫链蒙特卡罗技术，并且似乎比基于完全似然的方法更适用于更大的数据集。
