Multi-class cancer classification via partial least squares with gene expression profiles

MOTIVATION: Discrimination between two classes such as normal and cancer samples and between two types of cancers based on gene expression profiles is an important problem which has practical implications as well as the potential to further our understanding of gene expression of various cancer cells. Classification or discrimination of more than two groups or classes (multi-class) is also needed. The need for multi-class discrimination methodologies is apparent in many microarray experiments where various cancer types are considered simultaneously. RESULTS: Thus, in this paper we present the extension to the classification methodology proposed earlier Nguyen and Rocke (2002b; Bioinformatics, 18, 39-50) to classify cancer samples from multiple classes. The methodologies proposed in this paper are applied to four gene expression data sets with multiple classes: (a) a hereditary breast cancer data set with (1) BRCA1-mutation, (2) BRCA2-mutation and (3) sporadic breast cancer samples, (b) an acute leukemia data set with (1) acute myeloid leukemia (AML), (2) T-cell acute lymphoblastic leukemia (T-ALL) and (3) B-cell acute lymphoblastic leukemia (B-ALL) samples, (c) a lymphoma data set with (1) diffuse large B-cell lymphoma (DLBCL), (2) B-cell chronic lymphocytic leukemia (BCLL) and (3) follicular lymphoma (FL) samples, and (d) the NCI60 data set with cell lines derived from cancers of various sites of origin. In addition, we evaluated the classification algorithms and examined the variability of the error rates using simulations based on randomization of the real data sets. We note that there are other methods for addressing multi-class prediction recently and our approach is along the line of Nguyen and Rocke (2002b; Bioinformatics, 18, 39-50). CONTACT: dnguyen@stat.tamu.edu; dmrocke@ucdavis.edu     

Tumor classification by partial least squares using microarray gene expression data

MOTIVATION: One important application of gene expression microarray data is classification of samples into categories, such as the type of tumor. The use of microarrays allows simultaneous monitoring of thousands of genes expressions per sample. This ability to measure gene expression en masse has resulted in data with the number of variables p(genes) far exceeding the number of samples N. Standard statistical methodologies in classification and prediction do not work well or even at all when N < p. Modification of existing statistical methodologies or development of new methodologies is needed for the analysis of microarray data. RESULTS: We propose a novel analysis procedure for classifying (predicting) human tumor samples based on microarray gene expressions. This procedure involves dimension reduction using Partial Least Squares (PLS) and classification using Logistic Discrimination (LD) and Quadratic Discriminant Analysis (QDA). We compare PLS to the well known dimension reduction method of Principal Components Analysis (PCA). Under many circumstances PLS proves superior; we illustrate a condition when PCA particularly fails to predict well relative to PLS. The proposed methods were applied to five different microarray data sets involving various human tumor samples: (1) normal versus ovarian tumor; (2) Acute Myeloid Leukemia (AML) versus Acute Lymphoblastic Leukemia (ALL); (3) Diffuse Large B-cell Lymphoma (DLBCLL) versus B-cell Chronic Lymphocytic Leukemia (BCLL); (4) normal versus colon tumor; and (5) Non-Small-Cell-Lung-Carcinoma (NSCLC) versus renal samples. Stability of classification results and methods were further assessed by re-randomization studies.     

样条变换偏最小二乘在肝癌数据分类中的应用

肝癌是中国最常见的恶性肿瘤之一。基于肿瘤基因表达谱数据的分析与研究是当今研究的热点,对于癌症的早期诊断、治疗具有十分重要的意义。针对高维小样本基因表达谱数据所显现的变量间严重共线性、类别变量与预测变量的非线性关系,采用了基于样条变换的偏最小二乘回归新技术。首先通过筛选法去除基因表达谱数据中的冗余信息,然后以3次B基样条变换实现非线性基因表达谱数据的线性化重构,随后将重构的矩阵交由偏最小二乘法构建类别变量与预测变量间的关系模型。最后,通过对肝癌肿瘤基因表达谱数据的分析,结果显示此分类模型对数据重构稳健,有效的解决了高维小样本基因表达谱数据间的过拟合和变量间的共线性,具有较高的拟合和分类正确率。     

Classification using partial least squares with penalized logistic regression

MOTIVATION: One important aspect of data-mining of microarray data is to discover the molecular variation among cancers. In microarray studies, the number n of samples is relatively small compared to the number p of genes per sample (usually in thousands). It is known that standard statistical methods in classification are efficient (i.e. in the present case, yield successful classifiers) particularly when n is (far) larger than p. This naturally calls for the use of a dimension reduction procedure together with the classification one. RESULTS: In this paper, the question of classification in such a high-dimensional setting is addressed. We view the classification problem as a regression one with few observations and many predictor variables. We propose a new method combining partial least squares (PLS) and Ridge penalized logistic regression. We review the existing methods based on PLS and/or penalized likelihood techniques, outline their interest in some cases and theoretically explain their sometimes poor behavior. Our procedure is compared with these other classifiers. The predictive performance of the resulting classification rule is illustrated on three data sets: Leukemia, Colon and Prostate.     

Linking gene expression data with patient survival times using partial least squares

There is an increasing need to link the large amount of genotypic data, gathered using microarrays for example, with various phenotypic data from patients. The classification problem in which gene expression data serve as predictors and a class label phenotype as the binary outcome variable has been examined extensively, but there has been less emphasis in dealing with other types of phenotypic data. In particular, patient survival times with censoring are often not used directly as a response variable due to the complications that arise from censoring. We show that the issues involving censored data can be circumvented by reformulating the problem as a standard Poisson regression problem. The procedure for solving the transformed problem is a combination of two approaches: partial least squares, a regression technique that is especially effective when there is severe collinearity due to a large number of predictors, and generalized linear regression, which extends standard linear regression to deal with various types of response variables. The linear combinations of the original variables identified by the method are highly correlated with the patient survival times and at the same time account for the variability in the covariates. The algorithm is fast, as it does not involve any matrix decompositions in the iterations. We apply our method to data sets from lung carcinoma and diffuse large B-cell lymphoma studies to verify its effectiveness.     

Delaunay triangulation with partial least squares projection to latent structures: a model for G-protein coupled receptors classification and fast structure recognition

As an important transmembrane protein family in eukaryon, G-protein coupled receptors (GPCRs) play a significant role in cellular signal transduction and are important targets for drug design. However, it is very difficult to resolve their tertiary structure by X-ray crystallography. In this study, we have developed a Delaunay model, which constructs a series of simplexes with latent variables to classify the families of GPCRs and projects unknown sequences to principle component space (PC-space) to predict their topology. Computational results show that, for the classification of GPCRs, the method achieves the accuracy of 91.0 and 87.6% for Class A, more than 80% for the other three classes in differentiating GPCRs from non-GPCRs and 70% for discriminating between four major classes of GPCR, respectively. When recognizing the structure of GPCRs, all the N-terminals of sequences can be determined correctly. The maximum accuracy of predicting transmembrane segments is achieved in the 7th transmembrane segment of Rhodopsin, which is 99.4%, and the average error is 2.1 amino acids, which is the lowest in all of the segments prediction. This method could provide structural information of a novel GPCR as a tool for experiments and other algorithms of structure prediction of GPCRs. Academic users should send their request for the MATLAB program for classifying GPCRs and predicting the topology of them at liml@scu.edu.cn .     

Avian skull morphological evolution: exploring exo- and endocranial covariation with two-block partial least squares

While rostral variation has been the subject of detailed avian evolutionary research, avian skull organization, characterized by a flexed or extended appearance of the skull, has eventually become neglected by mainstream evolutionary inquiries. This study aims to recapture its significance, evaluating possible functional, phylogenetic and developmental factors that may be underlying it. In order to estimate which, and how, elements of the skull intervene in patterning the skull we tested the statistical interplay between a series of old mid-sagittal angular measurements (mostly endocranial) in combination with newly obtained skull metrics based on landmark superimposition methods (exclusively exocranial shape), by means of the statistic-morphometric technique of two-block partial least squares. As classic literature anticipated, we found that the external appearance of the skull corresponds to the way in which the plane of the caudal cranial base is oriented, in connection with the orientations of the plane of the foramen magnum and of the lateral semicircular canal. The pattern of covariation found between metrics conveys flexed or extended appearances of the skull implicitly within a single and statistically significant dimension of covariation. Marked shape changes with which angles covary concentrate at the supraoccipital bone, the cranial base and the antorbital window, whereas the plane measuring the orientation of the anterior portion of the rostrum does not intervene. Statistical covariance between elements of the caudal cranial base and the occiput inplies that morphological integration underlies avian skull macroevolutionary organization as a by-product of the regional concordance of such correlated elements within the early embryonic chordal domain of mesodermic origin.     

Predicting student performance via NAEP secondary art analysis using partial least squares SEM

This article describes a secondary analysis of the National Assessment of Educational Progress 2008 eighth-grade visual arts data (N = 3,912). These assessments occur under government mandate on a periodic schedule and data on the arts were collected in 1997 and again in 2008. The purpose of this study was to predict students' visual art response performance using students' home environment, personal characteristics, in-school curriculum, and art-related not-for-school (extracurricular) activities. Formative measurement models and structural paths were modeled in structural equation modeling (SEM) using Smart PLS. The initial SEM model included four latent constructs and one endogenous variable measuring students' performance. Both direct and indirect effects between latent constructs were modeled and assessed. Altogether, the four latent constructs explained 21.3% of the variance students' responding performance, out of which home environment construct had the strongest impact. School-related artistic activities in school do predict students' performance significantly but in lesser strength. Students' personal attributes and their art-related not-for-school activities predict students' performance to a substantially lesser degree. Implications of these findings will be discussed in terms of the data findings and larger issues of what these data represent as a means of following curriculum articulation with standards and the impact of art specialists in schools.     

Iterative partial least squares with right-censored data analysis: a comparison to other dimension reduction techniques

In the linear model with right-censored responses and many potential explanatory variables, regression parameter estimates may be unstable or, when the covariates outnumber the uncensored observations, not estimable. We propose an iterative algorithm for partial least squares, based on the Buckley-James estimating equation, to estimate the covariate effect and predict the response for a future subject with a given set of covariates. We use a leave-two-out cross-validation method for empirically selecting the number of components in the partial least-squares fit that approximately minimizes the error in estimating the covariate effect of a future observation. Simulation studies compare the methods discussed here with other dimension reduction techniques. Data from the AIDS Clinical Trials Group protocol 333 are used to motivate the methodology.     

A comparison of neural networks and partial least squares for deconvoluting fluorescence spectra

This article compares backpropagation neural networks (BNN) with partial least squares (PLS) techniques in terms of their ability to deconvolute fluorescence spectra. Both actual experimental and simulated spectral data are studied for 2 binary systems. These systems consist of mixtures of tryptophan and tyrosine, and NADH and tryptophan over a total concentration range of 10(-7) to 10(-4) M. It is shown that BNN is superior to PLS for both systems.     

Hemodialysis monitoring using mid‐ and near‐infrared spectroscopy with partial least squares regression

Blood constituents such as urea, glucose, lactate, phosphate and creatinine are of high relevance in monitoring the process of detoxification in ambulant dialysis treatment. In the present work, 2 different vibrational spectroscopic techniques are used to determine those molecules quantitatively in artificial dialysate solutions. The goal of the study is to compare the performance of near‐infrared (NIR) and mid‐infrared (MIR) spectroscopy in hyphenation with partial least squares regression (PLSR) directly by using the same sample set. The results show that MIR spectroscopy is better suited to analyze the analytes of interest. Multilevel multifactor design is used to cover the relevant concentration variations during dialysis. MIR spectroscopy coupled to a multi reflection attenuated total reflection (ATR) cell enables reliable prediction of all target analytes. In contrast, the NIR spectroscopic method does not give access to all 5 components but only to urea and glucose. For both methods, coefficients of determination greater or equal to 0.86 can be achieved in the test‐set validation process for urea and glucose. Lactate, phosphate and creatinine perform well in the MIR with R² ≥ 0.95 using test‐set validation.      

Reconstruction of genetic association networks from microarray data: a partial least squares approach

MOTIVATION: Gene association/interaction networks provide vast amounts of information about essential processes inside the cell. A complete picture of gene-gene associations/interactions would open new horizons for biologists, ranging from pure appreciation to successful manipulation of biological pathways for therapeutic purposes. Therefore, identification of important biological complexes whose members (genes and their products proteins) interact with each other is of prime importance. Numerous experimental methods exist but, for the most part, they are costly and labor intensive. Computational techniques, such as the one proposed in this work, provide a quick 'budget' solution that can be used as a screening tool before more expensive techniques are attempted. Here, we introduce a novel computational method based on the partial least squares (PLS) regression technique for reconstruction of genetic networks from microarray data. RESULTS: The proposed PLS method is shown to be an effective screening procedure for the detection of gene-gene interactions from microarray data. Both simulated and real microarray experiments show that the PLS-based approach is superior to its competitors both in terms of performance and applicability. AVAILABILITY: R code is available from the supplementary web-site whose URL is given below.     

Surrogate variable analysis using partial least squares (SVA-PLS) in gene expression studies

MOTIVATION: In a typical gene expression profiling study, our prime objective is to identify the genes that are differentially expressed between the samples from two different tissue types. Commonly, standard analysis of variance (ANOVA)/regression is implemented to identify the relative effects of these genes over the two types of samples from their respective arrays of expression levels. But, this technique becomes fundamentally flawed when there are unaccounted sources of variability in these arrays (latent variables attributable to different biological, environmental or other factors relevant in the context). These factors distort the true picture of differential gene expression between the two tissue types and introduce spurious signals of expression heterogeneity. As a result, many genes which are actually differentially expressed are not detected, whereas many others are falsely identified as positives. Moreover, these distortions can be different for different genes. Thus, it is also not possible to get rid of these variations by simple array normalizations. This both-way error can lead to a serious loss in sensitivity and specificity, thereby causing a severe inefficiency in the underlying multiple testing problem. In this work, we attempt to identify the hidden effects of the underlying latent factors in a gene expression profiling study by partial least squares (PLS) and apply ANCOVA technique with the PLS-identified signatures of these hidden effects as covariates, in order to identify the genes that are truly differentially expressed between the two concerned tissue types. RESULTS: We compare the performance of our method SVA-PLS with standard ANOVA and a relatively recent technique of surrogate variable analysis (SVA), on a wide variety of simulation settings (incorporating different effects of the hidden variable, under situations with varying signal intensities and gene groupings). In all settings, our method yields the highest sensitivity while maintaining relatively reasonable values for the specificity, false discovery rate and false non-discovery rate. Application of our method to gene expression profiling for acute megakaryoblastic leukemia shows that our method detects an additional six genes, that are missed by both the standard ANOVA method as well as SVA, but may be relevant to this disease, as can be seen from mining the existing literature.     

Analysis of genetic marker-phenotype relationships by jack-knifed partial least squares regression (PLSR)

The utility of a relatively new multivariate method, bi-linear modelling by cross-validated partial least squares regression (PLSR), was investigated in the analysis of QTL. The distinguishing feature of PLSR is to reveal reliable covariance structures in data of different types with regard to the same set objects. Two matrices X (here: genetic markers) and Y (here: phenotypes) are interactively decomposed into latent variables (PLS components, or PCs) in a way which facilitates statistically reliable and graphically interpretable model building. Natural collinearities between input variables are utilized actively to stabilise the modelling, instead of being treated as a statistical problem. The importance of cross-validation/jack-knifing as an intuitively appealing way to avoid overfitting, is emphasized. Two datasets from chromosomal mapping studies of different complexity were chosen for illustration (QTL for tomato yield and for oat heading date). Results from PLSR analysis were compared to published results and to results using the package PLABQTL in these data sets. In all cases PLSR gave at least similar explained validation variances as the reported studies. An attractive feature is that PLSR allows the analysis of several traits/replicates in one analysis, and the direct visual identification of individuals with desirable marker genotypes. It is suggested that PLSR may be useful in structural and functional genomics and in marker assisted selection, particularly in cases with limited number of objects.     

Use of partial least squares regression to impute SNP genotypes in Italian Cattle breeds

Background

The objective of the present study was to test the ability of the partial least squares regression technique to impute genotypes from low density single nucleotide polymorphisms (SNP) panels i.e. 3K or 7K to a high density panel with 50K SNP. No pedigree information was used.

Methods

Data consisted of 2093 Holstein, 749 Brown Swiss and 479 Simmental bulls genotyped with the Illumina 50K Beadchip. First, a single-breed approach was applied by using only data from Holstein animals. Then, to enlarge the training population, data from the three breeds were combined and a multi-breed analysis was performed. Accuracies of genotypes imputed using the partial least squares regression method were compared with those obtained by using the Beagle software. The impact of genotype imputation on breeding value prediction was evaluated for milk yield, fat content and protein content.

Results

In the single-breed approach, the accuracy of imputation using partial least squares regression was around 90 and 94% for the 3K and 7K platforms, respectively; corresponding accuracies obtained with Beagle were around 85% and 90%. Moreover, computing time required by the partial least squares regression method was on average around 10 times lower than computing time required by Beagle. Using the partial least squares regression method in the multi-breed resulted in lower imputation accuracies than using single-breed data. The impact of the SNP-genotype imputation on the accuracy of direct genomic breeding values was small. The correlation between estimates of genetic merit obtained by using imputed versus actual genotypes was around 0.96 for the 7K chip.

Conclusions

Results of the present work suggested that the partial least squares regression imputation method could be useful to impute SNP genotypes when pedigree information is not available.