Dimensionality reduction for visualizing high-dimensional biological data

Malepathirana, Tamasha; Senanayake, Damith; Vidanaarachchi, Rajith; Gautam, Vini; Halgamuge, Saman K.

doi:10.1016/j.biosystems.2022.104749

Cited by 6 publications

(7 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Classification Due to the increase in high-dimensional data and the limited number of samples, the "big P small n" paradigm has become a major challenge in the field of biomedical data mining 1,2 . Especially for microarray profile datasets, the number of genes is much larger than the number of samples, but only a few feature genes are closely related to cancer 3,4 . Feature selection can remove irrelevant and redundant genes, improve the classification and diagnosis rate of cancer, and help to improve the treatment of cancer 5,6 .…”

Section: Accmentioning

confidence: 99%

A new hybrid algorithm for three-stage gene selection based on whale optimization

Liu

Zhang

et al. 2023

Sci Rep

View full text Add to dashboard Cite

In biomedical data mining, the gene dimension is often much larger than the sample size. To solve this problem, we need to use a feature selection algorithm to select feature gene subsets with a strong correlation with phenotype to ensure the accuracy of subsequent analysis. This paper presents a new three-stage hybrid feature gene selection method, that combines a variance filter, extremely randomized tree, and whale optimization algorithm. First, a variance filter is used to reduce the dimension of the feature gene space, and an extremely randomized tree is used to further reduce the feature gene set. Finally, the whale optimization algorithm is used to select the optimal feature gene subset. We evaluate the proposed method with three different classifiers in seven published gene expression profile datasets and compare it with other advanced feature selection algorithms. The results show that the proposed method has significant advantages in a variety of evaluation indicators.

show abstract

Section: Accmentioning

confidence: 99%

A new hybrid algorithm for three-stage gene selection based on whale optimization

Liu

Zhang

et al. 2023

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Metatranscriptomics in its turn Molecular data resulting from transcriptomics, proteomics, metabolomics, epigenomics and microbiomics/metagenomics assays are highly dimensional, implying that the number of measured molecular features, that is, genes or metabolites, greatly outnumbers the number of observations, that is, samples or patients. This is referred to as the curse of dimensionality [18]. Additionally, data encounter sparsity at several levels.…”

Section: Different Data Modalities Synergistically Define the Neurolo...mentioning

confidence: 99%

“…genes or metabolites measured greatly outnumbers the number of observations. This is referred to as the curse of dimensionality 18 . Additionally, data encounter sparsity at several levels.…”

Section: Different Data Modalities Synergistically Define the Neurolo...mentioning

confidence: 99%

“…This implicates that data points in the reduced dimension matrix are actually linear combinations of the original data points, so that they lose their one-on-one relationship with individual molecular features. Other methods such as Principal Coordinate analysis (PCoA), UMAP or t-SNE have been specifically optimized for high-dimensional and sparse biological data and can be used to create intuitive visualizations of underlying data clusters, trajectories and patterns of interest 18 . PCoA is commonly used to visualize microbiome beta-diversity due to its intrinsic visualization of dissimilarity between samples and robust performance on highly sparse data.…”

Section: Computational Strategies For Biomarker Discovery and Patient...mentioning

confidence: 99%

“…This implies that data points in the reduced dimension matrix are actually linear combinations of the original data points, so that they lose their one-on-one relationship with individual molecular features. Other methods such as principal coordinate analysis (PCoA), UMAP (Uniform Manifold Approximation and Projection) or t-SNE (t-distributed stochastic neighbor embedding) have been specifically optimized for highdimensional and sparse biological data and can be used to create intuitive visualizations of underlying data clusters, trajectories and patterns of interest[18]. PCoA is commonly used to visualize microbiome beta-diversity due to its intrinsic visualization of dissimilarity between samples and robust performance on very sparse data.Using a human discovery cohort, microbiomics and PCoA, Liang and colleagues found significant microbiome differences in individuals characterized by mild or no cognitive impairment, which were mainly caused by species belonging to the Firmicutes, Bacteriodetes and Proteobacteria phyla[22].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Molecular systems biology approaches to investigate mechanisms of gut−brain communication in neurological diseases

Boris

Vermeirssen

2023

Euro J of Neurology

View full text Add to dashboard Cite

BackgroundWhilst the incidence of neurological diseases is increasing worldwide, treatment remains mostly limited to symptom management. The gut−brain axis, which encompasses the communication routes between microbiota, gut and brain, has emerged as a crucial area of investigation for identifying new preventive and therapeutic targets in neurological disease.MethodsDue to the inter‐organ, systemic nature of the gut−brain axis, together with the multitude of biomolecules and microbial species involved, molecular systems biology approaches are required to accurately investigate the mechanisms of gut−brain communication. High‐throughput omics profiling, together with computational methodologies such as dimensionality reduction or clustering, machine learning, network inference and genome‐scale metabolic models, allows novel biomarkers to be discovered and elucidates mechanistic insights.ResultsIn this review, the general concepts of experimental and computational methodologies for gut−brain axis research are introduced and their applications are discussed, mainly in human cohorts. Important aspects are further highlighted concerning rational study design, sampling procedures and data modalities relevant for gut−brain communication, strengths and limitations of methodological approaches and some future perspectives.ConclusionMulti‐omics analyses, together with advanced data mining, are essential to functionally characterize the gut−brain axis and put forward novel preventive or therapeutic strategies in neurological disease.

show abstract

Motor intent recognition of multi-feature fusion EEG signals by UMAP algorithm

Sui

Wang³

et al. 2023

Med Biol Eng Comput

View full text Add to dashboard Cite

Dimensionality reduction for visualizing high-dimensional biological data

Cited by 6 publications

References 76 publications

A new hybrid algorithm for three-stage gene selection based on whale optimization

A new hybrid algorithm for three-stage gene selection based on whale optimization

Molecular systems biology approaches to investigate mechanisms of gut−brain communication in neurological diseases

Motor intent recognition of multi-feature fusion EEG signals by UMAP algorithm

Contact Info

Product

Resources

About