Algorithms for Molecular Kinetics

科学通过其工具取得进步，而这项提案是对QUB继续发展的续订申请， 分子动力学分析软件工具包。集成的QUB软件包可以分析任何 由马尔可夫模型描述的系统。它运行在Windows下的Intel PC上，并提供了各种辅助 信号处理和报告生成工具。QUB还负责数据采集和信号调理。其核心是 从数据中求解速率常数的逆马尔可夫算法可以将数据理想化或对原始数据进行运算 数据。该拟合提供了参数的误差限制以及用于模型比较的对数似然度。《The Variable》 度量优化器使用似然函数的解析导数，并接受许多先验约束，包括 明细余额。例程可以应用于不同的刺激，例如不同的电压、膜张力或 求出配体浓度对刺激依赖的速率常数的贡献。理想化的数据是 可用于各种排序的平稳性或动态不均匀性等。数据库操作允许标绘 任何参数(幅度、概率、持续时间、标准差、突发长度、局部相似性等) 或记录中的时间或位置。提供了任意函数对所有数据的曲线拟合。自动程序设计 在Word和Excel中生成输出报告。QUB模型库包括分子马达的阶梯模型， 多通道同时活动模型的合并和优化，求解宏观动力学 (多个通道)以马尔可夫模型代替时间常数，并在单个通道上完成仿真 渠道和宏观层面。Python脚本工具允许用户自动化，并且提供了大量的用户帮助 提供在线命令和教程。我们建议继续开发和维护QUB。我们会 开发非理想马尔可夫数据的算法，指导模型识别，提高计算效率， 和非平稳刺激的设计与分析。我们将创建交互式向导来简化以下典型任务 初学者，继续教授QUB软件课程，并写一本关于用QUB做动力学的书。

Science advances through its tools, and this proposal is a renewal application to continue development of QuB, a software toolkit for molecular kinetic analysis. The integrated QuB software package can analyze the kinetics of any system that is described by a Markov model. It runs on Intel PCs under Windows, and provides a variety of auxiliary signal processing and report generating tools. QuB also handles data acquisition and signal conditioning. The core inverse Markov algorithms ¿ solving for rate constants from the data ¿ can idealize the data or operate on the raw data. The fitting provides error limits on the parameters as well as the logliklihood for model comparison. The varible metric optimizers use analytical derivatives of the likelihood function and accept many a priori constraints including detailed balance. The routines can be applied across varying stimuli such as different voltages, membrane tensions, or ligand concentrations to solve for the stimulus dependent contributions to the rate constants. The idealized data is available for a variety of sorting for stationarity or kinetic inhomogenity, etc. The database operations permit plotting any parameter (amplitude, probability, duration, standard deviations, burst length, local likihood etc) against any other or time or position in the record. Curve fitting of arbitrary functions to all data is provided. The programautomatically generates output reports in Word and Excel. The QuB model library includes staircase models for molecular motors, the merging and optimization of models for multiple simultaneous channel activity, solving macroscopic kinetics (multiple channels) in terms of Markov models instead of time constants, and complete simulation at the single channel and the macroscopic level. Python scripting tools allow for user automation, and extensive user help is provided with on-line commands and tutorials. We propose to continue QuB development and maintenance. We will develop algorithms for nonideal Markov data, guided model identification, improvement of computational efficiency, and non-stationary stimulus design and analysis. We will create interactive wizards for simplifying typical tasks for beginners, continue to teach the QuB software course, and write a book on doing kinetics with QuB.