ST‐PLS: a multi‐directional nearest shrunken centroid type classifier via PLS

Sæbø, Solve; Almøy, Trygve; Aarøe, Jørgen; Aastveit, Are H.

doi:10.1002/cem.1101

Cited by 36 publications

(25 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For detailed description of the covariance structure of our data, we use two measures analogous to Sæbø et al in [12]. The condition index, first used in [13], and the absolute value of the covariances between the principal components of and the response vector as used in [12].…”

Section: Methodsmentioning

confidence: 99%

“…The condition index, first used in [13], and the absolute value of the covariances between the principal components of and the response vector as used in [12]. The condition index is used as a measure for variable dependence, with being the kth eigenvalue of .…”

Section: Methodsmentioning

confidence: 99%

“…A rapid increase means, the features have a strong linear dependence, a weak increase implies a weak dependence. If we now consider the principal components, like in [12], the relevance of a component is measured by means of the absolute value of the covariances () between the principal component and the class vector . Here equals 1 if sample belongs to group , otherwise equals -1, .…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

An Extension of PPLS-DA for Classification and Comparison to Ordinary PLS-DA

et al. 2013

View full text Add to dashboard Cite

Classification studies are widely applied, e.g. in biomedical research to classify objects/patients into predefined groups. The goal is to find a classification function/rule which assigns each object/patient to a unique group with the greatest possible accuracy (classification error). Especially in gene expression experiments often a lot of variables (genes) are measured for only few objects/patients. A suitable approach is the well-known method PLS-DA, which searches for a transformation to a lower dimensional space. Resulting new components are linear combinations of the original variables. An advancement of PLS-DA leads to PPLS-DA, introducing a so called ‘power parameter’, which is maximized towards the correlation between the components and the group-membership. We introduce an extension of PPLS-DA for optimizing this power parameter towards the final aim, namely towards a minimal classification error. We compare this new extension with the original PPLS-DA and also with the ordinary PLS-DA using simulated and experimental datasets. For the investigated data sets with weak linear dependency between features/variables, no improvement is shown for PPLS-DA and for the extensions compared to PLS-DA. A very weak linear dependency, a low proportion of differentially expressed genes for simulated data, does not lead to an improvement of PPLS-DA over PLS-DA, but our extension shows a lower prediction error. On the contrary, for the data set with strong between-feature collinearity and a low proportion of differentially expressed genes and a large total number of genes, the prediction error of PPLS-DA and the extensions is clearly lower than for PLS-DA. Moreover we compare these prediction results with results of support vector machines with linear kernel and linear discriminant analysis.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

An Extension of PPLS-DA for Classification and Comparison to Ordinary PLS-DA

et al. 2013

View full text Add to dashboard Cite

show abstract

“…In a recent proof of concept study, we introduced multivariate analysis in the form of Soft-Thresholding Partial Least Squares (ST-PLS) [3] for the mapping of genotype and phenotype interactions in the yeast, Saccharomyces cerevisiae [4], which has been at the center point for this development. Multivariate approaches have the potential to provide superior statistical power, increased interpretability of results and a deeper functional understanding of the genotype-phenotype landscape as it pays attention to relationships between multiple genotypes and multiple phenotypes, without producing an excessive number of hypotheses to test.…”

Section: Introductionmentioning

confidence: 99%

Improving stability and understandability of genotype-phenotype mapping in Saccharomyces using regularized variable selection in L-PLS regression

et al. 2012

Self Cite

View full text Add to dashboard Cite

BackgroundMultivariate approaches have been successfully applied to genome wide association studies. Recently, a Partial Least Squares (PLS) based approach was introduced for mapping yeast genotype-phenotype relations, where background information such as gene function classification, gene dispensability, recent or ancient gene copy number variations and the presence of premature stop codons or frameshift mutations in reading frames, were used post hoc to explain selected genes. One of the latest advancement in PLS named L-Partial Least Squares (L-PLS), where ‘L’ presents the used data structure, enables the use of background information at the modeling level. Here, a modification of L-PLS with variable importance on projection (VIP) was implemented using a stepwise regularized procedure for gene and background information selection. Results were compared to PLS-based procedures, where no background information was used.ResultsApplying the proposed methodology to yeast Saccharomyces cerevisiae data, we found the relationship between genotype-phenotype to have improved understandability. Phenotypic variations were explained by the variations of relatively stable genes and stable background variations. The suggested procedure provides an automatic way for genotype-phenotype mapping. The selected phenotype influencing genes were evolving 29% faster than non-influential genes, and the current results are supported by a recently conducted study. Further power analysis on simulated data verified that the proposed methodology selects relevant variables.ConclusionsA modification of L-PLS with VIP in a stepwise regularized elimination procedure can improve the understandability and stability of selected genes and background information. The approach is recommended for genome wide association studies where background information is available.

show abstract

“…In [28], PLS is followed by classical discrimination, and compared to following PLS with regularized discriminant analysis (RDA) on near infrared (NIR) spectra and carbohydrates content data. In [29], the idea of shrunken centroid originally proposed by Tibshirani is introduced in the conventional PLS paradigm, with the goal of performing classification and variable selection simultaneously. In [30], a new classification strategy is proposed with the purpose to combining the strength of both PLS-DA and SIMCA.…”

Section: Introductionmentioning

confidence: 99%

Using OrPLS to identify asymptomatic women at risk for Alzheimer's disease

Rayens

Liu

Andersen

et al. 2008

Journal of Chemometrics

View full text Add to dashboard Cite

Persons at risk for Alzheimer's disease (AD) demonstrate altered cortical activation measured by functional MRI (fMRI) years before symptoms of disease are expected. We used fMRI to study the differences in cortical activation between 13 women with a family history of AD and at least one apolipoprotein E4 allele, a risk factor for AD, and a control group of 11 women lacking both factors. Our primary goal was to assess how well the two groups are able to be statistically separated, a task which directly affects the performance of post hoc classification. The dimension of the dataset, however, precludes the use of ordinary classification methods. In this paper we show the superiority of using Oriented Partial Least Squares (OrPLS) to accomplish the classification in the presence of this dimensionality problem. We are able to reduce the misclassification rates on the standardized fMRI data from an average of about 48% for PCA, to an average of 27% for PLS, and then to perfect classification for OrPLS. Copyright ß 2008 John Wiley & Sons, Ltd.Keywords: Alzheimer's disease; functional magnetic resonance imaging; partial least squares; discriminant analysis STATEMENT OF THE PROBLEMThe problem of distinguishing normal persons at risk of developing Alzheimer's disease, at a later time from other normal persons who will not, is intrinsically more difficult than distinguishing normal persons from patients with established dementia. This is because the underlying neuropathology is generally more advanced in persons with dementia than in persons who are normal but who will later become demented. This more advanced neuropathology causes greater alterations in the function of large-scale neuronal networks, and greater structural changes, e.g., loss of brain volume or atrophy, that are more easily detected using the imaging methods available today [1].Normal persons at risk for Alzheimer's disease (AD) demonstrate altered cortical activation measured by functional MRI (fMRI) years before symptoms of the disease are expected. We used fMRI to study the differences in cortical activation between 13 women with a family history of AD and at least one apolipoprotein E4 allele, a risk factor for AD, and a control group of 11 women lacking both factors. Our principal goal was to assess the ability of fMRI to function as a biomarker for diagnosis in the carrier and non-carrier groups. This will allow us to ascertain how well the two groups can be separated based on the fMRI signal, and additionally which regions seem to contribute most to this separation. This separation is important because identifying normal persons who have significant brain functional alterations implies that they already manifest early changes in neuronal network function, perhaps in compensation for subtle underlying AD neuropathology, even though these persons are still functioning and behaving normally. These persons represent the best candidates for preventive treatment aimed at altering the course of AD (disease-modifying treatment).The original study on this ...

show abstract

ST‐PLS: a multi‐directional nearest shrunken centroid type classifier via PLS

Cited by 36 publications

References 28 publications

An Extension of PPLS-DA for Classification and Comparison to Ordinary PLS-DA

An Extension of PPLS-DA for Classification and Comparison to Ordinary PLS-DA

Improving stability and understandability of genotype-phenotype mapping in Saccharomyces using regularized variable selection in L-PLS regression

Using OrPLS to identify asymptomatic women at risk for Alzheimer's disease

Contact Info

Product

Resources

About