High-throughput Mass Spectrometry Data Research Articles

LDA enhanced one-bit compressive sensing method for high-throughput mass spectrometry data feature selection

Mass spectrometry (MS) is commonly used for the proteins detection of biological samples. However, it generates high-throughput data, posing huge computational challenges for signal processing. In previous studies, we have proposed a one-bit perturbed block CS (OB-PBCS) method to realize feature selection (FS) and obtain a feature set with reduced dimension for classification. However, its feature selection is realized by solving the reconstruction problem with simple greedy strategies, which fails to consider the final classification objective during reconstruction. Considering such problem, in this paper, we propose a hybrid linear discriminant analysis (LDA) enhanced one-bit CS method, called L-OBPBCS, which selects features by considering the subsequent classification and consists of three parts. The first part roughly filters out the features that do not favor the classification task with the proposed single feature decision coefficient. Then, OB-PBCS obtains significant features from the filtered feature set. Afterwards, LDA algorithm is further used for refining, where the amplitude of the optimal projection vector is gathered as the linear discrimination mapping coefficients to identify important features. To show the efficiency of our proposed L-OBPBCS method, classification based on the selected features for both synthesized dataset and real ovarian cancer MS dataset is evaluated. Results show that the feature set derived by L-OBPBCS ensures excellent classification performance, which surpasses the existing one-bit type FS method, and the computational complexity of L-OBPBCS is obviously reduced compared with the existing OB-PBCS method.

Imputation of Missing Values for Multi-Biospecimen Metabolomics Studies: Bias and Effects on Statistical Validity.

The analysis of high-throughput metabolomics mass spectrometry data across multiple biological sample types (biospecimens) poses challenges due to missing data. During differential abundance analysis, dropping samples with missing values can lead to severe loss of data as well as biased results in group comparisons and effect size estimates. However, the imputation of missing data (the process of replacing missing data with estimated values such as a mean) may compromise the inherent intra-subject correlation of a metabolite across multiple biospecimens from the same subject, which in turn may compromise the efficacy of the statistical analysis of differential metabolites in biomarker discovery. We investigated imputation strategies when considering multiple biospecimens from the same subject. We compared a novel, but simple, approach that consists of combining the two biospecimen data matrices (rows and columns of subjects and metabolites) and imputes the two biospecimen data matrices together to an approach that imputes each biospecimen data matrix separately. We then compared the bias in the estimation of the intra-subject multi-specimen correlation and its effects on the validity of statistical significance tests between two approaches. The combined approach to multi-biospecimen studies has not been evaluated previously even though it is intuitive and easy to implement. We examine these two approaches for five imputation methods: random forest, k nearest neighbor, expectation-maximization with bootstrap, quantile regression, and half the minimum observed value. Combining the biospecimen data matrices for imputation did not greatly increase efficacy in conserving the correlation structure or improving accuracy in the statistical conclusions for most of the methods examined. Random forest tended to outperform the other methods in all performance metrics, except specificity.

DIAMetAlyzer allows automated false-discovery rate-controlled analysis for data-independent acquisition in metabolomics

The extraction of meaningful biological knowledge from high-throughput mass spectrometry data relies on limiting false discoveries to a manageable amount. For targeted approaches in metabolomics a main challenge is the detection of false positive metabolic features in the low signal-to-noise ranges of data-independent acquisition results and their filtering. Another factor is that the creation of assay libraries for data-independent acquisition analysis and the processing of extracted ion chromatograms have not been automated in metabolomics. Here we present a fully automated open-source workflow for high-throughput metabolomics that combines data-dependent and data-independent acquisition for library generation, analysis, and statistical validation, with rigorous control of the false-discovery rate while matching manual analysis regarding quantification accuracy. Using an experimentally specific data-dependent acquisition library based on reference substances allows for accurate identification of compounds and markers from data-independent acquisition data in low concentrations, facilitating biomarker quantification.

LDA Enhanced One-Bit Compressive Sensing Method for High-Throughput Mass Spectrometry Data Feature Selection

LDA Enhanced One-Bit Compressive Sensing Method for High-Throughput Mass Spectrometry Data Feature Selection

Comparison of imputation and imputation-free methods for statistical analysis of mass spectrometry data with missing data.

Missing values are common in high-throughput mass spectrometry data. Two strategies are available to address missing values: (i) eliminate or impute the missing values and apply statistical methods that require complete data and (ii) use statistical methods that specifically account for missing values without imputation (imputation-free methods). This study reviews the effect of sample size and percentage of missing values on statistical inference for multiple methods under these two strategies. With increasing missingness, the ability of imputation and imputation-free methods to identify differentially and non-differentially regulated compounds in a two-group comparison study declined. Random forest and k-nearest neighbor imputation combined with a Wilcoxon test performed well in statistical testing for up to 50% missingness with little bias in estimating the effect size. Quantile regression imputation accompanied with a Wilcoxon test also had good statistical testing outcomes but substantially distorted the difference in means between groups. None of the imputation-free methods performed consistently better for statistical testing than imputation methods.

Feature selection method based on support vector machine and shape analysis for high-throughput medical data

Proteomics data analysis based on the mass-spectrometry technique can provide a powerful tool for early diagnosis of tumors and other diseases. It can be used for exploring the features that reflect the difference between samples from high-throughput mass spectrometry data, which are important for the identification of tumor markers. Proteomics mass spectrometry data have the characteristics of too few samples, too many features and noise interference, which pose a great challenge to traditional machine learning methods. Traditional unsupervised dimensionality reduction methods do not utilize the label information effectively, so the subspaces they find may not be the most separable ones of the data. To overcome the shortcomings of traditional methods, in this paper, we present a novel feature selection method based on support vector machine (SVM) and shape analysis. In the process of feature selection, our method considers not only the interaction between features but also the relationship between features and class labels, which improves the classification performance. The experimental results obtained from four groups of proteomics data show that, compared with traditional unsupervised feature extraction methods (i.e., Principal Component Analysis - Procrustes Analysis, PCA-PA), our method not only ensures that fewer features are selected but also ensures a high recognition rate. In addition, compared with the two kinds of multivariate filter methods, i.e., Max-Relevance Min-Redundancy (MRMR) and Fast Correlation-Based Filter (FCBF), our method has a higher recognition rate.

Quantitative Shotgun Proteomics Unveils Candidate Novel Esophageal Adenocarcinoma (EAC)-specific Proteins

Esophageal cancer is the eighth most common cancer worldwide and the majority of patients have systemic disease at presentation. Esophageal adenocarcinoma (OAC), the predominant subtype in western countries, is largely resistant to current chemotherapy regimens. Selective markers are needed to enhance clinical staging and to allow targeted therapies yet there are minimal proteomic data on this cancer type. After histological review, lysates from OAC and matched normal esophageal and gastric samples from seven patients were subjected to LC MS/MS after tandem mass tag labeling and OFFGEL fractionation. Patient matched samples of OAC, normal esophagus, normal stomach, lymph node metastases and uninvolved lymph nodes were used from an additional 115 patients for verification of expression by immunohistochemistry (IHC).Over six thousand proteins were identified and quantified across samples. Quantitative reproducibility was excellent between technical replicates and a moderate correlation was seen across samples with the same histology. The quantitative accuracy was verified across the dynamic range for seven proteins by immunohistochemistry (IHC) on the originating tissues. Multiple novel tumor-specific candidates are proposed and EPCAM was verified by IHC.This shotgun proteomic study of OAC used a comparative quantitative approach to reveal proteins highly expressed in specific tissue types. Novel tumor-specific proteins are proposed and EPCAM was demonstrated to be specifically overexpressed in primary tumors and lymph node metastases compared with surrounding normal tissues. This candidate and others proposed in this study could be developed as tumor-specific targets for novel clinical staging and therapeutic approaches.

A Cautionary Tale on the Inclusion of Variable Posttranslational Modifications in Database-Dependent Searches of Mass Spectrometry Data.

Mass spectrometry-based proteomics allows in principle the identification of unknown target proteins of posttranslational modifications and the sites of attachment. Including a variety of posttranslational modifications in database-dependent searches of high-throughput mass spectrometry data holds the promise to gain spectrum assignments to modified peptides, thereby increasing the number of assigned spectra, and to identify potentially interesting modification events. However, these potential benefits come for the price of an increased search space, which can lead to reduced scores, increased score thresholds, and erroneous peptide spectrum matches. We have assessed here the advantages and disadvantages of including the variable posttranslational modifications methionine oxidation, protein N-terminal acetylation, cysteine carbamidomethylation, transformation of N-terminal glutamine to pyroglutamic acid (Gln→pyro-Glu), and deamidation of asparagine and glutamine. Based on calculations of local false discovery rates and comparisons to known features of the respective modifications, we recommend for searches of samples that were not enriched for specific posttranslational modifications to only include methionine oxidation, protein N-terminal acetylation, and peptide N-terminal Gln→pyro-Glu as variable modifications. The principle of the validation strategy adopted here can also be applied for assessing the inclusion of posttranslational modifications for differently prepared samples, or for additional modifications. In addition, we have reassessed the special properties of the ubiquitin footprint, which is the remainder of ubiquitin moieties attached to lysines after tryptic digest. We show here that the ubiquitin footprint often breaks off as neutral loss and that it can be distinguished from dicarbamidomethylation events.

Accounting for isotopic clustering in Fourier transform mass spectrometry data analysis for clinical diagnostic studies.

Mass spectrometry based clinical proteomics has emerged as a powerful tool for high-throughput protein profiling and biomarker discovery. Recent improvements in mass spectrometry technology have boosted the potential of proteomic studies in biomedical research. However, the complexity of the proteomic expression introduces new statistical challenges in summarizing and analyzing the acquired data. Statistical methods for optimally processing proteomic data are currently a growing field of research. In this paper we present simple, yet appropriate methods to preprocess, summarize and analyze high-throughput MALDI-FTICR mass spectrometry data, collected in a case-control fashion, while dealing with the statistical challenges that accompany such data. The known statistical properties of the isotopic distribution of the peptide molecules are used to preprocess the spectra and translate the proteomic expression into a condensed data set. Information on either the intensity level or the shape of the identified isotopic clusters is used to derive summary measures on which diagnostic rules for disease status allocation will be based. Results indicate that both the shape of the identified isotopic clusters and the overall intensity level carry information on the class outcome and can be used to predict the presence or absence of the disease.

Abstract PR13: Systems and structural biology approaches to elucidate new effectors in KRAS mutant tumors

Abstract Activated KRAS interacts directly and indirectly with numerous downstream effector pathways that are still elusive or poorly characterized. Furthermore, it has become clear over the past few years that the KRAS regulatory gene network (interactome) is highly cell-context dependent and thus tumor specific. We sought to computationally elucidate novel KRAS effectors, modulators, and interactors in Lung Adenocarcinomas (LUAD) and Colon Adenocarcinomas (COAD) by integrating computational and biochemical reverse engineering methods. Specifically, results from ARACNe and MINDy based analysis of KRAS networks were combined with Protein-protein interaction (PPI) predictions based on the PrePPI algorithm. These algorithms detect, respectively, transcriptional, post-translational, and protein-protein interactions respectively. The MARINa/VIPER algorithm was used to analyze the networks resulting from these analyses to identify key modulators and effectors of KRAS activity, including modulators harboring genetic alterations that phenocopy aberrant KRAS activity. In addition, we collaborated with the Jackson (Stanford) and Vidal (Harvard) labs to include high-throughput tandem affinity mass spectrometry data and yeast 2 hybrid screens in these analyses. Validation of computationally inferred effectors, using pooled shRNA silencing assays in primary lung cancer cells isolated from a KrasLSL/p53f/f mouse model of LUAD validated novel genes as likely KRASmut effectors and putative synthetic lethal genes. At least two hairpins against RNF43, E4F1, HMCN1, MYO9A, RHOT2, SETD1B and MAP3K2 scored in this assay. These results suggest that KRAS network analyses using computational methods can identify novel members of the KRAS interactome. Citation Format: Joshua Broyde, David Simpson, David A. Wah, Federico M. Giorgi, Donald Petrey, Mariano J. Alvarez, Antonina Silkov, Alexander Lachmann, David E. Hill, Marc Vidal, Peter Jackson, Barry Honig, Alejandro Sweet-Cordero, Andrea Califano. Systems and structural biology approaches to elucidate new effectors in KRAS mutant tumors. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr PR13.

Proteogenomic Analysis of Bradyrhizobium japonicum USDA110 Using Genosuite, an Automated Multi-algorithmic Pipeline

We present GenoSuite, an integrated proteogenomic pipeline to validate, refine and discover protein coding genes using high-throughput mass spectrometry (MS) data from prokaryotes. To demonstrate the effectiveness of GenoSuite, we analyzed proteomics data of Bradyrhizobium japonicum (USDA110), a model organism to study agriculturally important rhizobium-legume symbiosis. Our analysis confirmed 31% of known genes, refined 49 gene models for their translation initiation site (TIS) and discovered 59 novel protein coding genes. Notably, a novel protein which redefined the boundary of a crucial cytochrome P450 system related operon was discovered, known to be highly expressed in the anaerobic symbiotic bacteroids. A focused analysis on N-terminally acetylated peptides indicated downstream TIS for gene blr0594. Finally, ortho-proteogenomic analysis revealed three novel genes in recently sequenced B. japonicum USDA6(T) genome. The discovery of large number of missing genes and correction of gene models have expanded the proteomic landscape of B. japonicum and presents an unparalleled utility of proteogenomic analyses and versatility of GenoSuite for annotating prokaryotic genomes including pathogens.

Detecting outlier peptides in quantitative high-throughput mass spectrometry data

Quantitative high-throughput mass spectrometry has become an established tool to measure relative gene expression proteome-wide. The output of such an experiment usually consists of a list of expression ratios (fold changes) for several thousand proteins between two conditions. However, we observed that individual peptide fold changes may show a significantly different behavior than other peptides from the same protein and that these differences cannot be explained by imprecise measurements. Such outlier peptides can be the consequence of several technical (misidentifications, misquantifications) or biological (post-translational modifications, differential regulation of isoforms) reasons. We developed a method to detect outlier peptides in mass spectrometry data which is able to delineate imprecise measurements from real outlier peptides with high accuracy when the true difference is as small as 1.4 fold. We applied our method to experimental data and investigated the different technical and biological effects that result in outlier peptides. Our method will assist future research to reduce technical bias and can help to identify genes with differentially regulated protein isoforms in high throughput mass spectrometry data.

Mass Conservation and Inference of Metabolic Networks from High-Throughput Mass Spectrometry Data

We present a step towards the metabolome-wide computational inference of cellular metabolic reaction networks from metabolic profiling data, such as mass spectrometry. The reconstruction is based on identification of irreducible statistical interactions among the metabolite activities using the ARACNE reverse-engineering algorithm and on constraining possible metabolic transformations to satisfy the conservation of mass. The resulting algorithms are validated on synthetic data from an abridged computational model of Escherichia coli metabolism. Precision rates upwards of 50% are routinely observed for identification of full metabolic reactions, and recalls upwards of 20% are also seen.

A Consensus of Core Protein Complex Compositions for Saccharomyces cerevisiae

Analyses of biological processes would benefit from accurate definitions of protein complexes. High-throughput mass spectrometry data offer the possibility of systematically defining protein complexes; however, the predicted compositions vary substantially depending on the algorithm applied. We determine consensus compositions for 409 core protein complexes from Saccharomyces cerevisiae by merging previous predictions with a new approach. Various analyses indicate that the consensus is comprehensive and of high quality. For 85 out of 259 complexes not recorded in GO, literature search revealed strong support in the form of coprecipitation. New complexes were verified by an independent interaction assay and by gene expression profiling of strains with deleted subunits, often revealing which cellular processes are affected. The consensus complexes are available in various formats, including a merge with GO, resulting in 518 protein complex compositions. The utility is further demonstrated by comparison with binary interaction data to reveal interactions between core complexes.

Census for proteome quantification.

Quantitative analysis has become increasingly important in the proteomics field; however, the large amount of mass spectrometric data and the different types of quantitative strategies make data analysis ever challenging. Here we describe a quantitative software tool called Census to analyze high-throughput mass spectrometry data from shotgun proteomics experiments in an efficient way. Census is capable of analyzing various stable isotope labeling experiments (using, e.g., (15)N, (18)O, SILAC, iTRAQ, TMT) in addition to labeling-free experiments. With high-resolution data, Census increases the quantitative accuracy by minimizing the contributions of interfering peaks and chemical noise with a small accuracy tolerance for each isotope peak. Census provides various scoring algorithms including least-squares correlation, weight average, singleton peptide detection with discriminant analysis, and probability score for each peptide. Furthermore, Census has built-in multiple statistical filters to maintain robust quality control on quantitative results.

Classification-based comparison of pre-processing methods for interpretation of mass spectrometry generated clinical datasets

BackgroundMass spectrometry is increasingly being used to discover proteins or protein profiles associated with disease. Experimental design of mass-spectrometry studies has come under close scrutiny and the importance of strict protocols for sample collection is now understood. However, the question of how best to process the large quantities of data generated is still unanswered. Main challenges for the analysis are the choice of proper pre-processing and classification methods. While these two issues have been investigated in isolation, we propose to use the classification of patient samples as a clinically relevant benchmark for the evaluation of pre-processing methods.ResultsTwo in-house generated clinical SELDI-TOF MS datasets are used in this study as an example of high throughput mass-spectrometry data. We perform a systematic comparison of two commonly used pre-processing methods as implemented in Ciphergen ProteinChip Software and in the Cromwell package. With respect to reproducibility, Ciphergen and Cromwell pre-processing are largely comparable. We find that the overlap between peaks detected by either Ciphergen ProteinChip Software or Cromwell is large. This is especially the case for the more stringent peak detection settings. Moreover, similarity of the estimated intensities between matched peaks is high.We evaluate the pre-processing methods using five different classification methods. Classification is done in a double cross-validation protocol using repeated random sampling to obtain an unbiased estimate of classification accuracy. No pre-processing method significantly outperforms the other for all peak detection settings evaluated.ConclusionWe use classification of patient samples as a clinically relevant benchmark for the evaluation of pre-processing methods. Both pre-processing methods lead to similar classification results on an ovarian cancer and a Gaucher disease dataset. However, the settings for pre-processing parameters lead to large differences in classification accuracy and are therefore of crucial importance. We advocate the evaluation over a range of parameter settings when comparing pre-processing methods. Our analysis also demonstrates that reliable classification results can be obtained with a combination of strict sample handling and a well-defined classification protocol on clinical samples.

Computational prediction models for cancer classification using mass spectrometry data

Classification of human complex diseases such as cancers using high-throughput mass spectrometry data generated by modern proteomic technology has quickly become an attractive topic of research in bioinformatics. However, successful applications of such proteomic strategies for early disease detection are greatly dependent on the effectiveness of computational models for data analysis. Ultimately, the extraction of appropriate features that can represent the identities of different classes plays the frontal critical factor for any difficult classification problems. In addition, another major problem associated with pattern recognition is how to effectively handle a large feature space. This paper addresses these two frontal issues for Mass Spectrometry (MS) classification. We apply two computational prediction models to extract features of MS data and then use vector quantisation to reduce the feature storage. We also introduce the technique of information fusion for classification enhancement. The proposed methodology was tested using an MS-based ovarian cancer dataset and the results were found to be superior to a support vector machine approach using a different feature for the same data.

ProtQuant: a tool for the label-free quantification of MudPIT proteomics data

BackgroundEffective and economical methods for quantitative analysis of high throughput mass spectrometry data are essential to meet the goals of directly identifying, characterizing, and quantifying proteins from a particular cell state. Multidimensional Protein Identification Technology (MudPIT) is a common approach used in protein identification. Two types of methods are used to detect differential protein expression in MudPIT experiments: those involving stable isotope labelling and the so-called label-free methods. Label-free methods are based on the relationship between protein abundance and sampling statistics such as peptide count, spectral count, probabilistic peptide identification scores, and sum of peptide Sequest XCorr scores (ΣXCorr). Although a number of label-free methods for protein quantification have been described in the literature, there are few publicly available tools that implement these methods. We describe ProtQuant, a Java-based tool for label-free protein quantification that uses the previously published ΣXCorr method for quantification and includes an improved method for handling missing data.ResultsProtQuant was designed for ease of use and portability for the bench scientist. It implements the ΣXCorr method for label free protein quantification from MudPIT datasets. ProtQuant has a graphical user interface, accepts multiple file formats, is not limited by the size of the input files, and can process any number of replicates and any number of treatments. In addition,ProtQuant implements a new method for dealing with missing values for peptide scores used for quantification. The new algorithm, called ΣXCorr*, uses "below threshold" peptide scores to provide meaningful non-zero values for missing data points. We demonstrate that ΣXCorr* produces an average reduction in false positive identifications of differential expression of 25% compared to ΣXCorr.ConclusionProtQuant is a tool for protein quantification built for multi-platform use with an intuitive user interface. ProtQuant efficiently and uniquely performs label-free quantification of protein datasets produced with Sequest and provides the user with facilities for data management and analysis. Importantly, ProtQuant is available as a self-installing executable for the Windows environment used by many bench scientists.

Machine learning methods for predictive proteomics

The search for predictive biomarkers of disease from high-throughput mass spectrometry (MS) data requires a complex analysis path. Preprocessing and machine-learning modules are pipelined, starting from raw spectra, to set up a predictive classifier based on a shortlist of candidate features. As a machine-learning problem, proteomic profiling on MS data needs caution like the microarray case. The risk of overfitting and of selection bias effects is pervasive: not only potential features easily outnumber samples by 10(3) times, but it is easy to neglect information-leakage effects during preprocessing from spectra to peaks. The aim of this review is to explain how to build a general purpose design analysis protocol (DAP) for predictive proteomic profiling: we show how to limit leakage due to parameter tuning and how to organize classification and ranking on large numbers of replicate versions of the original data to avoid selection bias. The DAP can be used with alternative components, i.e. with different preprocessing methods (peak clustering or wavelet based), classifiers e.g. Support Vector Machine (SVM) or feature ranking methods (recursive feature elimination or I-Relief). A procedure for assessing stability and predictive value of the resulting biomarkers' list is also provided. The approach is exemplified with experiments on synthetic datasets (from the Cromwell MS simulator) and with publicly available datasets from cancer studies.

154: Integration of TPSA and High-Throughput Mass Spectrometry Data Improves Prostate Cancer Prediction

154: Integration of TPSA and High-Throughput Mass Spectrometry Data Improves Prostate Cancer Prediction