Data-dependent Methods Research Articles

Data-Based Choice of the Number of Pilot Stages for Plug-in Bandwidth Selection

The choice of the bandwidth is a crucial issue for kernel density estimation. Among all the data-dependent methods for choosing the bandwidth, the direct plug-in method has shown a particularly good performance in practice. This procedure is based on estimating an asymptotic approximation of the optimal bandwidth, using two “pilot” kernel estimation stages. Although two pilot stages seem to be enough for most densities, for a long time the problem of how to choose an appropriate number of stages has remained open. Here we propose an automatic (i.e., data-based) method for choosing the number of stages to be employed in the plug-in bandwidth selector. Asymptotic properties of the method are presented and an extensive simulation study is carried out to compare its small-sample performance with that of the most recommended bandwidth selectors in the literature.

Maximum Variance Hashing via Column Generation

With the explosive growth of the data volume in modern applications such as web search and multimedia retrieval, hashing is becoming increasingly important for efficient nearest neighbor (similar item) search. Recently, a number of data-dependent methods have been developed, reflecting the great potential of learning for hashing. Inspired by the classic nonlinear dimensionality reduction algorithm—maximum variance unfolding, we propose a novel unsupervised hashing method, named maximum variance hashing, in this work. The idea is to maximize the total variance of the hash codes while preserving the local structure of the training data. To solve the derived optimization problem, we propose a column generation algorithm, which directly learns the binary-valued hash functions. We then extend it using anchor graphs to reduce the computational cost. Experiments on large-scale image datasets demonstrate that the proposed method outperforms state-of-the-art hashing methods in many cases.

A Probabilistic Framework for Peptide and Protein Quantification from Data-Dependent and Data-Independent LC-MS Proteomics Experiments

A probability-based quantification framework is presented for the calculation of relative peptide and protein abundance in label-free and label-dependent LC-MS proteomics data. The results are accompanied by credible intervals and regulation probabilities. The algorithm takes into account data uncertainties via Poisson statistics modified by a noise contribution that is determined automatically during an initial normalization stage. Protein quantification relies on assignments of component peptides to the acquired data. These assignments are generally of variable reliability and may not be present across all of the experiments comprising an analysis. It is also possible for a peptide to be identified to more than one protein in a given mixture. For these reasons the algorithm accepts a prior probability of peptide assignment for each intensity measurement. The model is constructed in such a way that outliers of any type can be automatically reweighted. Two discrete normalization methods can be employed. The first method is based on a user-defined subset of peptides, while the second method relies on the presence of a dominant background of endogenous peptides for which the concentration is assumed to be unaffected. Normalization is performed using the same computational and statistical procedures employed by the main quantification algorithm. The performance of the algorithm will be illustrated on example data sets, and its utility demonstrated for typical proteomics applications. The quantification algorithm supports relative protein quantification based on precursor and product ion intensities acquired by means of data-dependent methods, originating from all common isotopically-labeled approaches, as well as label-free ion intensity-based data-independent methods.

Optimized Fast and Sensitive Acquisition Methods for Shotgun Proteomics on a Quadrupole Orbitrap Mass Spectrometer

Advances in proteomics are continually driven by the introduction of new mass spectrometric instrumentation with improved performances. The recently introduced quadrupole Orbitrap (Q Exactive) tandem mass spectrometer allows fast acquisition of high-resolution higher-energy collisional dissociation (HCD) tandem mass spectra due to the parallel mode of operation, where the generation, filling, and storage of fragment ions can be performed while simultaneously measuring another ion packet in the Orbitrap mass analyzer. In this study, data-dependent acquisition methods for "fast" or "sensitive" scanning were optimized and assessed by comparing stable isotope labeled yeast proteome coverage. We discovered that speed was the most important parameter for sample loads above 125 ng, where a 95 ms HCD scanning method allowed for identification and quantification of more than 2000 yeast proteins from 1 h of analysis time. At sample loads below 125 ng, a 156 ms HCD acquisition method improved the sensitivity, mass accuracy, and quality of data and enabled us to identify 30% more proteins and peptides than the faster scanning method. A similar effect was observed when the LC gradient was extended to 2 or 3 h for the analysis of complex mammalian whole cell lysates. Using a 3 h LC gradient, the sensitive method enabled identification of more than 4000 proteins from 1 μg of tryptic HeLa digest, which was almost 200 more identifications compared to the faster scanning method. Our results demonstrate that peptide identification on a quadrupole Orbitrap is dependent on sample amounts, acquisition speed, and data quality, which emphasizes the need for acquisition methods tailored for different sample loads and analytical preferences.

Analysis of plant galactolipids by reversed-phase high-performance liquid chromatography/mass spectrometry with accurate mass measurement

The composition of plant membrane lipids was investigated by reversed-phase high performance liquid chromatography mass spectrometry with accurate mass measurement. The data dependent methods for the analysis of monogalactosyldiacylglycerols (MGDGs) and digalactosyldiacylglycerols (DGDGs) have been developed. The optimised chromatographic systems were based on a 2.0mm i.d. Nucleosil C18 column with methanol/water (MGDGs) or acetonitrile/methanol/water (DGDGs) gradients. The galactolipids were ionised by electrospray operated in the positive ion mode and identified based on their MS/MS spectra. High resolution spectra with accurate masses were found to be essential for correct interpretation of the MS data. The elution order of non-oxidised MGDGs and DGDGs followed the equivalent carbon numbers. The methods were applied for detailed characterisation of the MGDGs and DGDGs in the leaves of Arabidopsis thaliana and Melissa officinalis.

Lag Length Selection for Unit Root Tests in the Presence of Nonstationary Volatility

A number of recently published papers have focused on the problem of testing for a unit root in the case where the driving shocks may be unconditionally heteroskedastic. These papers have, however, assumed that the lag length in the unit root test regression is a deterministic function of the sample size, rather than data-determined, the latter being standard empirical practice. In this paper we investigate the finite sample impact of unconditional heteroskedasticity on conventional data-dependent methods of lag selection in augmented Dickey-Fuller type unit root test regressions and propose new lag selection criteria which allow for the presence of heteroskedasticity in the shocks. We show that standard lag selection methods show a tendency to over-fit the lag order under heteroskedasticity, which results in significant power losses in the (wild bootstrap implementation of the) augmented Dickey-Fuller tests under the alternative. The new lag selection criteria we propose are shown to avoid this problem yet deliver unit roots with almost identical finite sample size and power properties as the corresponding tests based on conventional lag selection methods when the shocks are homoskedastic.

A Framework for Intelligent Data Acquisition and Real-Time Database Searching for Shotgun Proteomics

In the analysis of complex peptide mixtures by MS-based proteomics, many more peptides elute at any given time than can be identified and quantified by the mass spectrometer. This makes it desirable to optimally allocate peptide sequencing and narrow mass range quantification events. In computer science, intelligent agents are frequently used to make autonomous decisions in complex environments. Here we develop and describe a framework for intelligent data acquisition and real-time database searching and showcase selected examples. The intelligent agent is implemented in the MaxQuant computational proteomics environment, termed MaxQuant Real-Time. It analyzes data as it is acquired on the mass spectrometer, constructs isotope patterns and SILAC pair information as well as controls MS and tandem MS events based on real-time and prior MS data or external knowledge. Re-implementing a top10 method in the intelligent agent yields similar performance to the data dependent methods running on the mass spectrometer itself. We demonstrate the capabilities of MaxQuant Real-Time by creating a real-time search engine capable of identifying peptides “on-the-fly” within 30 ms, well within the time constraints of a shotgun fragmentation “topN” method. The agent can focus sequencing events onto peptides of specific interest, such as those originating from a specific gene ontology (GO) term, or peptides that are likely modified versions of already identified peptides. Finally, we demonstrate enhanced quantification of SILAC pairs whose ratios were poorly defined in survey spectra. MaxQuant Real-Time is flexible and can be applied to a large number of scenarios that would benefit from intelligent, directed data acquisition. Our framework should be especially useful for new instrument types, such as the quadrupole-Orbitrap, that are currently becoming available.

Automated Iterative MS/MS Acquisition: A Tool for Improving Efficiency of Protein Identification Using a LC–MALDI MS Workflow

We have developed an information-dependent, iterative MS/MS acquisition (IMMA) tool for improving MS/MS efficiency, increasing proteome coverage, and shortening analysis time for high-throughput proteomics applications based on the LC-MALDI MS/MS platform. The underlying principle of IMMA is to limit MS/MS analyses to a subset of molecular ions that are likely to identify a maximum number of proteins. IMMA reduces redundancy of MS/MS analyses by excluding from the precursor ion peak lists proteotypic peptides derived from the already identified proteins and uses a retention time prediction algorithm to limit the degree of false exclusions. It also increases the utilization rate of MS/MS spectra by removing "low value" unidentifiable targets like nonpeptides and peptides carrying large loads of modifications, which are flagged by their "nonpeptide" excess-to-nominal mass ratios. For some samples, IMMA increases the number of identified proteins by ∼20-40% when compared to the data dependent methods. IMMA terminates an MS/MS run at the operator-defined point when "costs" (e.g., time of analysis) start to overrun "benefits" (e.g., number of identified proteins), without prior knowledge of sample contents and complexity. To facilitate analysis of closely related samples, IMMA's inclusion list functionality is currently under development.

Isotopic Signature Transfer and Mass Pattern Prediction (IsoStamp): An Enabling Technique for Chemically-Directed Proteomics

Directed proteomics applies mass spectrometry analysis to a subset of information-rich proteins. Here we describe a method for targeting select proteins by chemical modification with a tag that imparts a distinct isotopic signature detectable in a full-scan mass spectrum. Termed isotopic signature transfer and mass pattern prediction (IsoStamp), the technique exploits the perturbing effects of a dibrominated chemical tag on a peptide’s mass envelope, which can be detected with high sensitivity and fidelity using a computational method. Applying IsoStamp, we were able to detect femtomole quantities of a single tagged protein from total mammalian cell lysates at signal-to-noise ratios as low as 2.5:1. To identify a tagged-peptide’s sequence, we performed an inclusion list-driven shotgun proteomics experiment where peptides bearing a recoded mass envelope were targeted for fragmentation, allowing for direct site mapping. Using this approach, femtomole quantities of several targeted peptides were identified in total mammalian cell lysate, while traditional data-dependent methods were unable to identify as many peptides. Additionally, the isotopic signature imparted by the dibromide tag was detectable on a 12-kDa protein, suggesting applications in identifying large peptide fragments, such as those containing multiple or large posttranslational modifications (e.g., glycosylation). IsoStamp has the potential to enhance any proteomics platform that employs chemical labeling for targeted protein identification, including isotope coded affinity tagging, isobaric tagging for relative and absolute quantitation, and chemical tagging strategies for posttranslational modification.

Glycation Isotopic Labeling with 13C-Reducing Sugars for Quantitative Analysis of Glycated Proteins in Human Plasma

Non-enzymatic glycation of proteins is a post-translational modification produced by a reaction between reducing sugars and amino groups located in lysine and arginine residues or in the N-terminal position. This modification plays a relevant role in medicine and food industry. In the clinical field, this undesired role is directly linked to blood glucose concentration and therefore to pathological conditions derived from hyperglycemia (>11 mm glucose) such as diabetes mellitus or renal failure. An approach for qualitative and quantitative analysis of glycated proteins is here proposed to achieve the three information levels for their complete characterization. These are: 1) identification of glycated proteins, 2) elucidation of sugar attachment sites, and 3) quantitative analysis to compare glycemic states. Qualitative analysis was carried out by tandem mass spectrometry after endoproteinase Glu-C digestion and boronate affinity chromatography for isolation of glycated peptides. For this purpose, two MS operational modes were used: higher energy collisional dissociation-MS2 and CID-MS3 by neutral loss scan monitoring of two selective neutral losses (162.05 and 84.04 Da for the glucose cleavage and an intermediate rearrangement of the glucose moiety). On the other hand, quantitative analysis was based on labeling of proteins with [(13)C(6)]glucose incubation to evaluate the native glycated proteins labeled with [(12)C(6)]glucose. As glycation is chemoselective, it is exclusively occurring in potential targets for in vivo modifications. This approach, named glycation isotopic labeling, enabled differentiation of glycated peptides labeled with both isotopic forms resulting from enzymatic digestion by mass spectrometry (6-Da mass shift/glycation site). The strategy was then applied to a reference plasma sample, revealing the detection of 50 glycated proteins and 161 sugar attachment positions with identification of preferential glycation sites for each protein. A predictive approach was also tested to detect potential glycation sites under high glucose concentration.

Quantitative Clinical Proteomics by Liquid Chromatography–Tandem Mass Spectrometry: Assessing the Platform

Clinical laboratories use electrospray ionization and mass spectrometry (MS)1 every day to measure dozens of different small molecule analytes. The capacity to precisely quantify these biomarkers has improved patient care. To build on the successful implementation of novel MS methodologies, many clinical mass spectrometrists hope to expand the list of measurable analytes to include peptides and proteins, several examples of which have been reported and reviewed (1). One can also envision the translation of novel biomarkers from the research-grade MS assays, with which they are discovered, to clinically acceptable MS assays for diagnosis, prognosis, and evaluation of therapeutic efficacy. An argument can be made that investigators should use clinically acceptable MS assays from the outset (2), which would facilitate the transition from research/discovery to clinical practice. Given the successes over the last few years in precisely measuring proteins by MS in clinically relevant matrices, this outcome may not be too much to ask. Traditionally, the word proteomics has encompassed efforts aimed at simultaneously measuring many proteins in a given complex sample. Shotgun proteomic experiments, which attempt to identify all of the proteins in a complex sample, and targeted proteomic experiments, which attempt to quantify a subset of the proteins known to exist in a complex sample, can both be carried out using liquid chromatography–tandem MS (LC-MS/MS). Shotgun proteomics is already being used clinically to identify the proteinaceous component of amyloid deposits (3). For the quantification of multiple protein concentrations in a complex sample, however, targeted methods are more precise because data-dependent shotgun proteomic methods variably sample fewer ions across chromatographic peaks. For this reason, targeted approaches are much more likely to be employed in the quantification of proteins in clinical specimens. Targeted methods typically use proteolysis with trypsin to digest proteins into peptides, which are more amenable to analysis …

Directed sample interrogation utilizing an accurate mass exclusion-based data-dependent acquisition strategy (AMEx).

The ability to perform thorough sampling is of critical importance when using mass spectrometry to characterize complex proteomic mixtures. A common approach is to reinterrogate a sample multiple times by LC-MS/MS. However, the conventional data-dependent acquisition methods that are typically used in proteomics studies will often redundantly sample high-intensity precursor ions while failing to sample low-intensity precursors entirely. We describe a method wherein the masses of successfully identified peptides are used to generate an accurate mass exclusion list such that those precursors are not selected for sequencing during subsequent analyses. We performed multiple concatenated analytical runs to sample a complex cell lysate, using either accurate mass exclusion-based data-dependent acquisition (AMEx) or standard data-dependent acquisition, and found that utilization of AMEx on an ESI-Orbitrap instrument significantly increases the total number of validated peptide identifications relative to a standard DDA approach. The additional identified peptides represent precursor ions that exhibit low signal intensity in the sample. Increasing the total number of peptide identifications augmented the number of proteins identified, as well as improved the sequence coverage of those proteins. Together, these data indicate that using AMEx is an effective strategy to improve the characterization of complex proteomic mixtures.

MRM screening/biomarker discovery with linear ion trap MS: a library of human cancer-specific peptides

BackgroundThe discovery of novel protein biomarkers is essential in the clinical setting to enable early disease diagnosis and increase survivability rates. To facilitate differential expression analysis and biomarker discovery, a variety of tandem mass spectrometry (MS/MS)-based protein profiling techniques have been developed. For achieving sensitive detection and accurate quantitation, targeted MS screening approaches, such as multiple reaction monitoring (MRM), have been implemented.MethodsMCF-7 breast cancer protein cellular extracts were analyzed by 2D-strong cation exchange (SCX)/reversed phase liquid chromatography (RPLC) separations interfaced to linear ion trap MS detection. MS data were interpreted with the Sequest-based Bioworks software (Thermo Electron). In-house developed Perl-scripts were used to calculate the spectral counts and the representative fragment ions for each peptide.ResultsIn this work, we report on the generation of a library of 9,677 peptides (p < 0.001), representing ~1,572 proteins from human breast cancer cells, that can be used for MRM/MS-based biomarker screening studies. For each protein, the library provides the number and sequence of detectable peptides, the charge state, the spectral count, the molecular weight, the parameters that characterize the quality of the tandem mass spectrum (p-value, DeltaM, Xcorr, DeltaCn, Sp, no. of matching a, b, y ions in the spectrum), the retention time, and the top 10 most intense product ions that correspond to a given peptide. Only proteins identified by at least two spectral counts are listed. The experimental distribution of protein frequencies, as a function of molecular weight, closely matched the theoretical distribution of proteins in the human proteome, as provided in the SwissProt database. The amino acid sequence coverage of the identified proteins ranged from 0.04% to 98.3%. The highest-abundance proteins in the cellular extract had a molecular weight (MW)<50,000.ConclusionPreliminary experiments have demonstrated that putative biomarkers, that are not detectable by conventional data dependent MS acquisition methods in complex un-fractionated samples, can be reliable identified with the information provided in this library. Based on the spectral count, the quality of a tandem mass spectrum and the m/z values for a parent peptide and its most abundant daughter ions, MRM conditions can be selected to enable the detection of target peptides and proteins.

Database searching and accounting of multiplexed precursor and product ion spectra from the data independent analysis of simple and complex peptide mixtures

A novel database search algorithm is presented for the qualitative identification of proteins over a wide dynamic range, both in simple and complex biological samples. The algorithm has been designed for the analysis of data originating from data independent acquisitions, whereby multiple precursor ions are fragmented simultaneously. Measurements used by the algorithm include retention time, ion intensities, charge state, and accurate masses on both precursor and product ions from LC-MS data. The search algorithm uses an iterative process whereby each iteration incrementally increases the selectivity, specificity, and sensitivity of the overall strategy. Increased specificity is obtained by utilizing a subset database search approach, whereby for each subsequent stage of the search, only those peptides from securely identified proteins are queried. Tentative peptide and protein identifications are ranked and scored by their relative correlation to a number of models of known and empirically derived physicochemical attributes of proteins and peptides. In addition, the algorithm utilizes decoy database techniques for automatically determining the false positive identification rates. The search algorithm has been tested by comparing the search results from a four-protein mixture, the same four-protein mixture spiked into a complex biological background, and a variety of other "system" type protein digest mixtures. The method was validated independently by data dependent methods, while concurrently relying on replication and selectivity. Comparisons were also performed with other commercially and publicly available peptide fragmentation search algorithms. The presented results demonstrate the ability to correctly identify peptides and proteins from data independent acquisition strategies with high sensitivity and specificity. They also illustrate a more comprehensive analysis of the samples studied; providing approximately 20% more protein identifications, compared to a more conventional data directed approach using the same identification criteria, with a concurrent increase in both sequence coverage and the number of modified peptides.

Challenges and strategies for targeted phosphorylation site identification and quantification using mass spectrometry analysis

Despite its importance, the 'ultimate' method to identify and quantify site-specific protein phosphorylation using mass spectrometry (MS) has yet to be established. This is as much a function of the dynamic range of instrumentation as it is the complexities surrounding the isolation and behavior of phosphopeptides. Phosphorylation site analysis using MS can be quite challenging when analyzing just one protein and quickly becomes a daunting task when attempting to perform proteome-wide measurements. Data-dependent tandem MS-based methods which are useful for the discovery and characterization of novel phosphorylation sites often lack the dynamic range and quantitative aspect required for studying the temporal phases of phosphorylation. While targeted methods such as multiple reaction monitoring do provide a highly specific and quantitative methodology for studying phosphorylation changes over time, they are not suited for initial discovery of previously unreported sites of phosphorylation. Data-independent acquisition represents a relatively new approach for simultaneous qualitative and quantitative sample analysis which holds promise for filling this technological gap.

Data dependency in multiple classifier systems

In this paper, the data dependency of aggregation modules in multiple classifier system is being investigated. We first propose a new categorization scheme, in which combining methods are grouped into data-independent, implicitly data-dependent and explicitly data-dependent. It is argued that data-dependent approaches present the highest potential for improved performance. In this study, we intend to provide a comprehensive investigation of this argument and explore the impact of data dependency on the performance of multiple classifiers. We evaluate this impact based on two criteria, prediction accuracy and stability. In addition, we examine the effect of class imbalance and uneven data distribution on these two criteria. This paper presents the findings of an extensive set of comparative experiments. Based on the findings, it can be concluded that data-dependent aggregation methods are generally more stable and less sensitive to class imbalance. In addition, data-dependent methods exhibited superior or identical generalization ability for most of the data sets.

Formula omitted]-Treatment ridit analyses: Resampling permutation methods

A g-treatment extension of the classical two-treatment ridit analysis is introduced. Concerns regarding distributional assumptions for ridit analyses are avoided by employing data dependent resampling methods. Upper-tail P -values based on resampling permutations are obtained for the g-treatment ridit analyses.

A microcapillary trap cartridge-microcapillary high-performance liquid chromatography electrospray ionization emitter device capable of peptide tandem mass spectrometry at the attomole level on an ion trap mass spectrometer with automated routine operation.

Reversed-phase microcapillary chromatography (RP-microLC) combined with electrospray ionization tandem mass spectrometry (ESI-MS/MS) is one of two prevailing techniques in proteomic analysis, the other being matrix-assisted laser desorption/ionization (MALDI). Despite the arguably better dynamic range obtainable with ESI, MALDI is increasingly popular due to ease of use, ruggedness and the ability to decouple separation from ionization. By contrast, in order to take advantage of the sensitivity and dynamic range afforded by the concentration-dependent nature of ESI, it is directly coupled to separations that take place in small i.d. RP-microLC columns. This gain in sensitivity often comes at a loss of ruggedness due to clogging of the small i.d. RP-microLC columns, one result of which is limited sample throughput. Here we describe a combined micropre-column-microLC-ESI device that is sensitive, rugged and modular in design allowing facile construction and troubleshooting. Due to low signal-to-noise as little as 1 attomole of a peptide can be selected by data-dependent methods for collision-induced dissociation. Importantly, the resulting tandem mass spectrum is of high enough quality to identify the peptide sequence by a database search against a complex database using SEQUEST. Finally, the device is demonstrated to be rugged as judged by >60 consecutive reversed-phase microLC separations on complex peptide mixtures before chromatographic resolution is degraded.

Data-dependent analyses in psychological research.

A data-dependent analysis assumes that all the information available to a researcher is contained within the observed data. Data-dependent methods for the analysis of experimental designs are shown to provide significant advantages over conventional techniques such as an F test. Two versions of three data-dependent methods based on permutations of the data are described and compared. One version utilizes ordinary least squares regression, and the other version utilizes least absolute deviations regression to analyze experimental designs. Analyses of an unbalanced two-way experimental design illustrate the differences among the six data-dependent approaches and the classical ordinary least squares F test, which depends on the assumptions of normality, homogeneity, and independence.

Probability binning and testing agreement between multivariate immunofluorescence histograms: extending the chi-squared test.

A key problem in immunohistochemistry is assessing when two sample histograms are significantly different. One test that is commonly used for this purpose in the univariate case is the chi-squared test. Comparing multivariate distributions is qualitatively harder, as the "curse of dimensionality" means that the number of bins can grow exponentially. For the chi-squared test to be useful, data-dependent binning methods must be employed. An example of how this can be done is provided by the "probability binning" method of Roederer et al. (1,2,3). We derive the theoretical distribution of the probability binning statistic, giving it a more rigorous foundation. We show that the null distribution is a scaled chi-square, and show how it can be related to the standard chi-squared statistic. A small simulation shows how the theoretical results can be used to (a) modify the probability binning statistic to make it more sensitive and (b) suggest variant statistics which, while still exploiting the data-dependent strengths of the probability binning procedure, may be easier to work with. The probability binning procedure effectively uses adaptive binning to locate structure in high-dimensional data. The derivation of a theoretical basis provides a more detailed interpretation of its behavior and renders the probability binning method more flexible.