Bidirectional movement of actin filaments along tracks of myosin heads

It is well established that muscle contraction results from the relative sliding of actin and myosin filaments. Both filaments have definite polarities and well-ordered structures. Thick filaments, however, are not vital for supporting movement in vitro. Previously we have demonstrated that actin filaments can move continuously on myosin fragments (subfragment-1 or heavy meromyosin (HMM] that are bound to a nitrocellulose surface. Here we report that actin filaments can move in opposite directions on tracks of myosin heads formed when actin filaments decorated with HMM are placed on a nitrocellulose surface. The actin filaments always move forward, frequently changing the direction of the movement, but never move backward reversing the polarity of the movement. The direction of movement is therefore determined by the polarity of the actin filament. These results indicate that myosin heads have considerable flexibility.     

The myosin motor in muscle generates a smaller and slower working stroke at higher load

Muscle contraction is driven by the motor protein myosin II, which binds transiently to an actin filament, generates a unitary filament displacement or 'working stroke', then detaches and repeats the cycle. The stroke size has been measured previously using isolated myosin II molecules at low load, with rather variable results, but not at the higher loads that the motor works against during muscle contraction. Here we used a novel X-ray-interference technique to measure the working stroke of myosin II at constant load in an intact muscle cell, preserving the native structure and function of the motor. We show that the stroke is smaller and slower at higher load. The stroke size at low load is likely to be set by a structural limit; at higher loads, the motor detaches from actin before reaching this limit. The load dependence of the myosin II stroke is the primary molecular determinant of the mechanical performance and efficiency of skeletal muscle.     

Myosin subfragment-1 is sufficient to move actin filaments in vitro

The rotating crossbridge model for muscle contraction proposes that force is produced by a change in angle of the crossbridge between the overlapping thick and thin filaments. Myosin, the major component of the thick filament, is comprised of two heavy chains and two pairs of light chains. Together they form two globular heads, which give rise to the crossbridge in muscle, and a coiled-coil rod, which forms the shaft of the thick filament. The isolated head fragment, subfragment-1 (S1), contains the ATPase and actin-binding activities of myosin (Fig. 1). Although S1 seems to have the requisite enzymatic activity, direct evidence that S1 is sufficient to drive actin movement has been lacking. It has long been recognized that in vitro movement assays are an important approach for identifying the elements in muscle responsible for force generation. Hynes et al. showed that beads coated with heavy meromyosin (HMM), a soluble proteolytic fragment of myosin consisting of a part of the rod and the two heads, can move on Nitella actin filaments. Using the myosin-coated surface assay of Kron and Spudich, Harada et al. showed that single-headed myosin filaments bound to glass support movement of actin at nearly the same speed as intact myosin filaments. These studies show that the terminal portion of the rod and the two-headed nature of myosin are not required for movement. To restrict the region responsible for movement further, we have modified the myosin-coated surface assay by replacing the glass surface with a nitrocellulose film. Here we report that myosin filaments, soluble myosin, HMM or S1, when bound to a nitrocellulose film, support actin sliding movement (Fig. 2). That S1 is sufficient to cause sliding movement of actin filaments in vitro gives strong support to models of contraction that place the site of active movement in muscle within the myosin head.     

一种新型汉语单音节识别方法

介绍了一种新型字基 ＶＱ／ＨＭＭ语音识别方法： ＶＱ与 ＨＭＭ分级识别算法。 使得 ＶＱ部分可用作语音识别的第一级处理， ＨＭＭ部分作第二级识别。在第一级识别中 可引进汉语的音素知识，使 ＶＱ／ＨＭＭ性能进一步提高，并可用于大字汇表的实时语音识 别，存贮量、计算量均大大减少。用此方法把汉语四声作一个 ＨＭＭ模型，使汉语 ４００ 个基本音节的识别率达 ９６％以上．若加上精确四声识别则可识别汉语的 １２００种声音。     

Two-headed binding of a processive myosin to F-actin

Myosins are motor proteins in cells. They move along actin by changing shape after making stereospecific interactions with the actin subunits. As these are arranged helically, a succession of steps will follow a helical path. However, if the myosin heads are long enough to span the actin helical repeat (approximately 36 nm), linear motion is possible. Muscle myosin (myosin II) heads are about 16 nm long, which is insufficient to span the repeat. Myosin V, however, has heads of about 31 nm that could span 36 nm and thus allow single two-headed molecules to transport cargo by walking straight. Here we use electron microscopy to show that while working, myosin V spans the helical repeat. The heads are mostly 13 actin subunits apart, with values of 11 or 15 also found. Typically the structure is polar and one head is curved, the other straighter. Single particle processing reveals the polarity of the underlying actin filament, showing that the curved head is the leading one. The shape of the leading head may correspond to the beginning of the working stroke of the motor. We also observe molecules attached by one head in this conformation.     

Sliding distance of actin filament induced by a myosin crossbridge during one ATP hydrolysis cycle

Muscle contraction results from a sliding movement of actin filaments induced by myosin crossbridges on hydrolysis of ATP, and many non-muscle cells are thought to move using a similar mechanism. The molecular mechanism of muscle contraction, however, is not completely understood. One of the major problems is the mechanochemical coupling at high velocity under near-zero load. Here, we report measurements of the sliding distance of an actin filament induced by a myosin crossbridge during one ATP hydrolysis cycle in an unloaded condition. We used single sarcomeres from which the Z-lines, structures which anchor the thin filaments in the sarcomere, had been completely removed by calcium-activated neutral protease (CANP) and trypsin, and measured both the sliding velocity of single actin filaments along myosin filaments and the ATPase activity during sliding. Our results show that the average sliding distance of the actin filament is less than or equal to 600 A during one ATP cycle, much longer than the length of power stroke of myosin crossbridges deduced from mechanical studies of muscle, which is of the order of 80 A (for example, ref. 15).     

基于隐马尔可夫模型的DNA序列识别

利用隐马尔可夫模型训练中不同结构的DNA序列的L值分布范围不同的特点,对传统多类投票模型进行改进,提出一种优于传统算法的快速训练算法,该算法只需训练出一类隐马尔可夫模型参数.对DNA内含子和外显子序列进行识别,平均识别率达到了90.8%.与支持向量机相比,隐马尔可夫模型在解决多分类问题方面具有优势,不但计算时间少,而且识别率高.     

Direct observation of the mechanochemical coupling in myosin Va during processive movement

Myosin Va transports intracellular cargoes along actin filaments in cells. This processive, two-headed motor takes multiple 36-nm steps in which the two heads swing forward alternately towards the barbed end of actin driven by ATP hydrolysis. The ability of myosin Va to move processively is a function of its long lever arm, the high duty ratio of its kinetic cycle and the gating of the kinetics between the two heads such that ADP release from the lead head is greatly retarded. Mechanical studies at the multiple- and the single-molecule level suggest that there is tight coupling (that is, one ATP is hydrolysed per power stroke), but this has not been directly demonstrated. We therefore investigated the coordination between the ATPase mechanism of the two heads of myosin Va and directly visualized the binding and dissociation of single fluorescently labelled nucleotide molecules, while simultaneously observing the stepping motion of the fluorescently labelled myosin Va as it moved along an actin filament. Here we show that preferential ADP dissociation from the trail head of mouse myosin Va is followed by ATP binding and a synchronous 36-nm step. Even at low ATP concentrations, the myosin Va molecule retained at least one nucleotide (ADP in the lead head position) when moving. Thus, we directly demonstrate tight coupling between myosin Va movement and the binding and dissociation of nucleotide by simultaneously imaging with near nanometre precision.     

具有分流叶片的离心式压缩机叶轮跨叶片流动计算

本文提出了计算具有分流叶片的离心式压缩机叶轮跨叶片方向的计算方法,这一方法运用统一的控制方程处理不同区域的问题,在每一个区域之间的数据结构联结均有程序自动完成。分流比的确定参照现有的方法,以分流叶片前缘点坐标插值获得,并考虑出口条件的影响。根据叶轮进口轴向分速和叶片对上游的影响确定上游过界条件,下游边界条件根据绝对速度等环量变化确定。算例结果表明,分流叶片长度、圆周位置均对流场有很大影响,当分流叶片前缘处于相邻两长叶片中间,而后部偏向长叶片压力面一侧时,流动状况有较大改善,并且叶片载荷减小,出口库塔条件满足。     

Hybrid Genetic Algorithm Based Optimization of Coupled HMM for Complex Interacting Processes Recognition

Coupled Hidden Markov Model (CHMM) is the extension of traditional HMM, which is mainly used for complex interactive process modeling such as two-hand gestures. However, the problems of finding optimal model parameter arc still of great interest to the researches in this area. This paper proposes a hybrid genetic algorithm (HGA) for the CHMM training. Chaos is used to initialize GA and used as mutation operator. Experiments on Chinese Tai‘Chi gestures show that standard GA (SC, A) based CHMM training is superior to Maximum Likelihood (ML) HMM training. HGA approach has the highest recognition rate of 98.0769%, then 96. 1538% for SGA. The last one is ML method, only with a recognition rate of 69.2308 %.     

参数共享在语音识别中的应用

参数共享是基于隐 Markov模型 (hidden Markovmodel,HMM)的语音识别系统的参数训练中的一个关键性问题 ,因此在语音识别的诸多领域中都有重要的应用。对参数共享的作用及其使用的聚类算法进行了分析研究 ,在此基础上提出改进合并分级聚类算法 ,并将其应用于 HMM的状态捆绑。实验表明 ,一个大规模词汇量的孤立词语音识别器采用 HMM的状态捆绑后 ,可以大大缩减识别过程的时空消耗 ,同时识别率仅有较小的损失     

HMM语音识别模型与一种修正训练算法

本文在统一的框架下描述了隐马尔柯夫模型（ＨＭＭ）用于语音识别时的各种形式，包括离散ＨＭＭ、连续混合密度ＨＭＭ、半连续ＨＭＭ和最大分量连续ＨＭＭ等，指出各种模型均是统一形式下的导出形式。文中就离散ＨＭＭ、连续混合密度ＨＭＭ和最大分量连续ＨＭＭ在非特定人全音节汉语语音识别中的应用，从识别率和复杂度两方面进行了性能比较。为提高最大分量连续ＨＭＭ的识别性能；提出了一种修正的训练算法。     

段长在汉语语音识别系统后处理阶段的应用

针对传统的隐含马尔可夫模型(hidden Markov model,HMM)存在的缺陷,该文提出了一种在识别的后处理阶段使用段长模型的方法,并应用在基于HMM的汉语识别系统上。该方法利用归一化的段长模型对识别系统的解码结果重新打分,比较前后两次算出的分数从而选出更可靠的识别结果。实验表明,通过该方法将段长模型应用在识别过程中,可以显著提高识别系统的性能,大量减少识别结果中的插入错误。数据显示,该方法使识别系统的音节错误率下降了大约10%,识别系统最终的插入错误和删除错误都低于1%。     

基于HMM模型的贝叶斯滤波方法

针对非线性动态系统估计问题,引入了一种不完全观测数据的统计模型--HMM模型,其双层随机结构较好地满足了实际估计的要求.在具体滤波过程中,将处理非线性问题常用的贝叶斯方法和HMM模型进行结合,通过预测和更新操作实现系统状态后验的连续递推估计,提高了运算速度.     

A lexicalized second-order-HMM for ambiguity resolution in Chinese segmentation and POS tagging

Hidden Markov Model（HMM） is a main solution to ambiguities in Chinese segmentation anti POS （part-of-speech） tagging. While most previous works tot HMM-based Chinese segmentation anti POS tagging eonsult POS informatiou in contexts, they do not utilize lexieal information which is crucial for resoMng certain morphologieal ambiguity. This paper proposes a method which incorporates lexieal information and wider context information into HMM. Model induction anti related smoothing technique are presented in detail. Experiments indicate that this technique improves the segmentation and tagging accuracy by nearly 1%.     

The Reconstructing of Low Signal-noise Ratio Single Ion Channel Signal from Patch-clamp Recordings Sampled in the Colored Background Noise

The single ion channel signal is an ionic current that can be recorded by the patch clamp technique. Hidden Markov model(HMM)algorithm has been used to convert the low signal-noise ra-tio (SNR) noisy recording into an idealized quantal one in the case of white background noise. The traditional HMM algorithm is extended and adapted to the colored background noise.A new algorithm called EHMM (Extended HMM) algorithm is proposed,and mainly validated by simulati-on.Results show that it's effective.     

Active fluidization of polymer networks through molecular motors

Entangled polymer solutions and melts exhibit elastic, solid-like resistance to quick deformations and a viscous, fluid-like response to slow deformations. This viscoelastic behaviour reflects the dynamics of individual polymer chains driven by brownian motion: since individual chains can only move in a snake-like fashion through the mesh of surrounding polymer molecules, their diffusive transport, described by reptation, is so slow that the relaxation of suddenly imposed stress is delayed. Entangled polymer solutions and melts therefore elastically resist deforming motions that occur faster than the stress relaxation time. Here we show that the protein myosin II permits active control over the viscoelastic behaviour of actin filament solutions. We find that when each actin filament in a polymerized actin solution interacts with at least one myosin minifilament, the stress relaxation time of the polymer solution is significantly shortened. We attribute this effect to myosin's action as a 'molecular motor', which allows it to interact with randomly oriented actin filaments and push them through the solution, thus enhancing longitudinal filament motion. By superseding reptation with sliding motion, the molecular motors thus overcome a fundamental principle of complex fluids: that only depolymerization makes an entangled, isotropic polymer solution fluid for quick deformations.     

有限态文法引导的基于连续密度HMM的连续汉语语音识别

介绍了用有限态文法引导的基于连续密度隐马尔可夫模型(HMM)的连续汉语语音识别系统.分析了系统的组成,词语的HMM的训练方法和对识别系统的测试结果。实验表明,用有限态文法辅助进行连续语音识别是一种有效的策略;连续密度HMM与基于矢量量化的HMM/VQ相比需要较大的计算量,但可明显提高识别准确率。     

基于HMM／MLFNN混合结构的说话人辨认研究

将隐马尔可夫模型与人工神经网络相结合既利用ＨＭＭ能够较好地描述动态时间序列又ＡＮＮ静态分类能力强的特点，应用于说话人辨认。本文将一个多层前馈神经网络与ＨＭＭ相结合构成混合模模型，与以往的方法不同。具有所需训练数据量小，推广性能良好的特点。