getPeaklist(IPPD)
getPeaklist()所属R语言包：IPPD
Peak pattern extraction by non-negative ls/lad template matching
非负LS /小伙子模板匹配的峰值特征提取
译者：生物统计家园网 机器人LoveR
描述----------Description----------
Generates a candidate list of isotopic peak patterns present in a protein mass spectrum. This is achieved by matching templates calculated according to the so-called Averagine model to the raw spectrum using either non-negative least squares (ls) or non-negative least absolute deviation (lad) estimation. The presence of multiple charge states is supported. In particular, the approach is capable of deconvolving overlapping patterns. The method can be applied with two different kind of peak shapes, Gaussians and Exponentially Modified Gaussians (EMG).
生成一个模式中的一种蛋白质质谱同位素峰的候选人名单。这是根据所谓的原始频谱使用任何非负最小二乘（LS）的非负最小绝对偏差（LAD）估计Averagine模型计算的模板匹配。支持多电荷态的存在。特别是，这种方法是，能够deconvolving重叠模式。该方法可应用于两个不同种类的峰形，指数修正的高斯和高斯（肌电图）。
用法----------Usage----------
getPeaklist(mz, intensities, model = c("Gaussian", "EMG"),
model.parameters = list(alpha = function(){},
sigma = function(){},
mu = function(){}),
loss = c("L2", "L1"), binning = FALSE,
postprocessing = TRUE, trace = TRUE, returnbasis = TRUE,
control.basis = list(charges = c(1,2,3,4),
eps = 1e-05),
control.localnoise = list(quantile = 0.5,
factor.place = 1.5,
factor.post = 0,
window = NULL,
subtract = FALSE),
control.postprocessing = list(mzfilter = FALSE,
prune = FALSE,
factor.prune = NULL,
ppm = NULL,
goodnessoffit = FALSE),
control.binning = list(tol = 0.01))
参数----------Arguments----------
参数：mz
A numeric vector of m/z (mass/charge) values (in Thomson), ordered increasingly.
一个numericm / z为（质量/电荷）值（汤姆逊）的向量，越来越有序。
参数：intensities
A numeric vector of intensities corresponding to mz.
一个numeric强度对应mz的向量。
参数：model
Basic model for the shape of a single peak. Must be "Gaussian" or "EMG" (exponentially modified Gaussian). See fitModelParameters for further information on these models.
基本模式为单峰形状。必须"Gaussian"或"EMG"（指数修改高斯）。对这些模型的进一步信息，请参阅fitModelParameters。
参数：loss
The loss function to be used. The choice loss = "L2" yield a nonnegative least squares fit, loss = "L1" a nonnegative least absolute deviation fit. The second choice is more robust when deviations from model assumptions (peak model, Averagine model,...) frequently occur in the data. Note, however, that computation time is much higher for the second choice (at least by a factor two).
损失函数可以使用。选择loss = "L2"产生一个非负最小二乘法拟合，loss = "L1"一个非负最小绝对偏差适合。第二种选择是更强大的模型假设（高峰模型，Averagine;模型，...）经常发生在数据的偏差。但是请注意，该计算时间的第二选择（至少有一个因素二）高得多。
参数：model.parameters
A list of functions with precisely one argument representing mz. The parameters of a single peak are typically modeled as a function of m/z. If model = "Gaussian", the peak shape depends on the parameter sigma (a function sigma(mz)). If model = "EMG", the peak shape additionally depends on two parameters alpha and mu (two functions alpha(mz) and mu(mz)). Note that constant functions are usually specified by using a construction of the form parameter(mz) <- function(mz) rep(constant, length(mz)). Moreover, note that a valid function has to be vectorized. For the automatic generation of those functions from a raw spectrum and further details on the meaning of the parameters, see fitModelParameters. The output of a call to fitModelParameters can directly be plugged into getPeaklist via the argument model.parameters.
正是一个代表mz的说法与列表功能。单峰的参数通常为蓝本的m / z的函数。 model = "Gaussian"如果，峰形参数取决于sigma（功能sigma(mz)）。如果model = "EMG"，峰形此外取决于两个参数alpha和mu（两个函数alpha(mz)和mu(mz)）。注意：常量函数通常指定使用形式parameter(mz) <- function(mz) rep(constant, length(mz))建设。此外，注意，有一个有效的功能进行量化。对于这些职能从原始频谱和参数的含义进一步的细节自动生成，看到fitModelParameters“。打检测到fitModelParameters输出可以直接插入到getPeaklist通过参数model.parameters。
参数：binning
A logical indicating whether the fitting process should be done sequentially in 'bins'. If TRUE, the spectrum is cut into pieces (bins). Each bin is then fitted separately, and the results of all bins are combined in the end. Division into bins may be configured using control.binning. See also the 'Details' section below.
一个logical指示是否在装修过程中应做“垃圾桶”的顺序。如果TRUE，频谱切成件（箱）。每个bin，然后分别安装，所有垃圾箱，结果在年底结合。成箱的司可配置使用control.binning。也见“详细资料”一节。
参数：postprocessing
A logical indicating whether a post-processing correction should be applied to the raw peaklist. See also the argument control.postprocessing and the 'Details' section below.
一个logical指示是否应适用于原料peaklist后处理校正。参见参数control.postprocessing和“详细资料”一节。
参数：trace
A logical indicating whether information tracing the different steps of the fitting process should be displayed.
一个logical指示是否应显示信息，跟踪装修过程中的不同步骤。
参数：returnbasis
A logical indicating whether the matrix of basis functions (template functions evaluated at mz) should be returned. Note that this may be expensive in terms of storage.
一个logical指示是否应退还基函数矩阵（mz模板功能评估）。请注意，这可能是昂贵的存储方面。
参数：control.basis
A list of arguments controlling the computation of the matrix of basis functions:
控制的基函数矩阵的计算参数列表：
chargesThe set of charge states present in the spectrum.
charges电荷态存在于频谱。
epsFunction values below eps are set equal to precisely zero in order to make the basis function matrix sparse.
eps函数值低于eps设置为等于精确为零，为了使基函数矩阵稀疏。
参数：control.localnoise
A list of arguments controlling the placement and selection of basis functions on the basis of a 'local noise level':
参数控制的基础功能上的一个地方的噪音水平的基础上的位置和选择列表：
quantileA value from 0.1, 0.2, ..., 0.9, specifying the quantile of the intensities residing in a sliding m/z window (s. below) to be used as 'local noise level'.
quantile从价值0.1, 0.2, ..., 0.9，指定居住在强度分量滑动的m / zwindow（S.以下）将用于为当地的噪音水平。
factor.placeControls the placement of basis functions. A basis function is placed at an element of mz if and only if the intensity at that position exceeds the 'local noise level' by a factor at least equal to factor.place.
factor.place控制的基础功能的位置。基础功能放在一个元素：mz如果只，如果在该位置的强度超过了当地的噪音水平的一个因素，至少等于factor.place。
factor.postControls which basis functions enter the postprocessing step. A basis function is discarded before the postprocessing step if its estimated amplitude does not exceed the factor.post times the 'local noise level'. By default factor.post = 0. The pre-filtering step before postprocessing is mainly done for computational speed-up, and factor.post = 0 is supposed to yield the qualitatively best solution, though it may take additional time.
factor.post控制的基础功能进入后处理步骤。基础功能被丢弃后处理步骤之前，如果其估计的幅度不超过当地的噪音水平“factor.post倍。默认情况下factor.post = 0。预过滤步骤前，后处理，主要是做的运算速度，和factor.post = 0应该产生质的最佳解决方案，但它可能需要额外的时间。
windowThe length of the sliding window used to compute quantile, to be specified in Thomson. By default, window is chosen in such a way that it equals the length of the support of an 'average' basis function for charge state one.
window用来计算quantile，汤姆森指定的滑动窗口的长度。默认情况下，window选择这样一种方式，它等于平均充电状态的基础上功能的支持长度。
subtractA logical indicating whether the 'local noise level' should be subtracted from the observed intensities. Setting subtract = TRUE is typically beneficial in the sense that fitting of noise is reduced.
subtractAlogical指示当地的噪音水平是否应减去从观测到的强度。设置subtract = TRUE通常是有益于降低噪声的拟合感。
参数：control.postprocessing
A list of arguments controlling the postprocessing step (provided postprocessing = TRUE):
一个控制后处理步骤的参数列表（提供postprocessing = TRUE）：
mzfilterSetting mzfilter = TRUE removes basis functions at positions where peak patterns are highly improbable to occur, thereby removing peaks from the list which are likely to be noise peaks. This filter is sometimes called 'peptide mass rule', see Zubarev et al. (1996): Accuracy Requirements for Peptide Characterization by Monoisotopic Molecular Measurements, Anal. Chem.,88,4060-4063.
mzfilter设置为mzfilter = TRUE删除峰模式是极不可能发生的位置的基础功能，从而消除从列表中，这很可能是噪声峰值的峰。该过滤器有时也被称为“肽质量规则”，见Zubarev等。 （1996年）：由单一同位素分子测量，肛门肽表征的精度要求。化学。,88,4060-4063。
prune, factor.pruneSetting prune = TRUE activates a crude scheme that removes low-intensity peaks (likely to be noise peaks), as frequently occurring in regions with extremely intense peaks. According to this scheme, a peak is removed from the peak list if its amplitude is less than factor.prune times the locally most intense amplitude, where factor.prune typically ranges from 0.01 to 0.1.
prune，factor.prune设置prune = TRUE激活了原油的计划，消除低强度峰（可能是噪声峰值），频繁发生的区域非常激烈峰。根据这项计划，一个高峰，从高峰名单中删除，如果其幅度比factor.prune本地最激烈的振幅，其中factor.prune通常范围从0.01到0.1 。
ppmA ppm (= parts per million) tolerance value within which basis functions at different m/z positions are considered to be merged, s. 'Details' below. By default, that value is computed from the spacing of the first two m/z positions.
ppm一个PPM（每百万件）内，在不同的m / z位置的基础功能，被认为是要合并的公差值。下面的“详细资料”。默认情况下，该值由前两米/ z位置的间距计算。
goodnessoffitA logical indicating whether a local goodness-of-fit adjustment of the signa-to-noise ratio should be computed. Yields usually more reliable evaluation of the detected patterns, but is computationally more demanding.
goodnessoffitAlogical善良的适合当地的签名信噪比的调整是否应计算。收益率通常更可靠的检测模式的评价，但计算更为苛刻。
参数：control.binning
Controls the division of the spectrum into bins (if binning = TRUE). Based on the 'local noise level' described in control.localnoise, if within a range of (1+tol) Thomson no further significant position occurs, a bin is closed, and a new one is not opened as long as a new significant position occurs..
控制频谱划分成箱（如果binning = TRUE）。基于描述当地的噪音水平，control.localnoise，如果范围内的(1+tol)汤姆森没有进一步的显着位置时，一个bin是封闭的，并作为一个新的，是不是只要打开一个新的发生重要地位......
Details
详情----------Details----------
值----------Value----------
An object of class peaklist.
对象类peaklist。
警告----------Warning----------
Although we have tried to choose default values expected to produce sensible results, the user should carefully
虽然我们试图选择默认值，预计将产生有意义的结果，用户应仔细
警告----------Warning----------
Depending on the length and the resolution of the raw spectrum, fitting the whole spectrum simultaneously as recommended is expensive from a computational point of view, and may take up to several
根据不同的长度和原始频谱的决议，装修全谱同时建议从计算的角度是昂贵的，可能需要几个
参见----------See Also----------
fitModelParameters, peaklist
fitModelParameters，peaklist
举例----------Examples----------
### load data[＃＃加载数据]
data(toyspectrum)
data(toyspectrumsolution)
mz <- toyspectrum[,"x"]
intensities <- toyspectrum[,"yyy"]
### select mz range[＃选择MZ范围]
filter <- mz >= 2800 & mz <= 3200
### Extract peak patterns with model = "Gaussian"[＃提取高峰的模式与模型=“高斯”]
sigmafun <- function (mz) -8.5e-07 * mz + 6.09e-10 * mz^2 + 0.00076
gausslist <- getPeaklist(mz = mz[filter], intensities = intensities[filter],
model = "Gaussian",
model.parameters = list(sigma = sigmafun,
alpha = function(mz){},
mu = function(mz){}),
control.localnoise = list(quantile = 0.5, factor.place = 3))
show(gausslist)
### threshold list at signal-to-noise ratio = 2[＃信号信噪比阈值列表= 2]
peaklist <- threshold(gausslist, threshold = 2)
### Extract peak patterns with model = "EMG" and loss = "L1"[＃提取高峰的模式与模型=“肌电图”和亏损=“母语”]
alpha0 <- function(mz) 0.00001875 * 0.5 * 4/3 * mz
sigma0 <- function(mz) 0.00001875 * 0.5 * mz
mu0 <- function(mz) return(rep(-0.06162891, length(mz)))
EMGlist <- getPeaklist(mz = mz[filter], intensities = intensities[filter],
model = "EMG", loss = "L1",
model.parameters = list(sigma = sigma0,
alpha = alpha0,
mu = mu0),
control.localnoise = list(quantile = 0.5, factor.place = 3))
show(EMGlist)
peaklist2 <- threshold(EMGlist, threshold = 2)
### plot results of the 1st list and compare vs. 'truth' [＃一日列表图的结果，并比较与“真理”]
### 'ground truth'[＃“地面真相”]
solution <- toyspectrumsolution[toyspectrumsolution[,1] >= 2800 & toyspectrumsolution[,1] <= 3200,]
visualize(gausslist, mz[filter], intensities[filter], lower = 3150, upper = 3170,
truth = TRUE,
signal = TRUE,
fitted = TRUE,
postprocessed = TRUE,
booktrue = as.matrix(toyspectrumsolution),
cutoff.eps = 0.2)
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-2-25 22:24:41
