Unsupervised Feature Extraction Applied to Bioinformatics

Taguchi, Y-h.

doi:10.1007/978-3-030-22456-1

Cited by 62 publications

(222 citation statements)

References 0 publications

Supporting

Mentioning

221

Contrasting

Order By: Relevance

“…In order to apply TD based unsupervised FE [15] to downloaded gene expression profiles, they must be formatted as a tensor. In this analysis, they are formatted as tensor, x ijkm ∈ R N ×24×18×2 , for N genes, 24 tissues, 18 drug treatments and two replicates.…”

Section: Td Based Unsupervised Fementioning

confidence: 99%

“…In this analysis, they are formatted as tensor, x ijkm ∈ R N ×24×18×2 , for N genes, 24 tissues, 18 drug treatments and two replicates. Then HOSVD [15] algorithm was applied to x ijkm and we get TD…”

Section: Td Based Unsupervised Fementioning

confidence: 99%

“…P i is corrected via BH criterion [16] and I, a set of genes i associated with adjusted P -values less than 0.01, is selected. For more detailed explanation of TD based unsupervised FE, see the recently published monograph [15].…”

Section: Td Based Unsupervised Fementioning

confidence: 99%

“…Although the number of drugs tested is as many as 15, because of additional three conditions, the total number of drug treatments is as many as 18. Usually, the first singular value vectors represent uniform values (i.e., the components that are not distinct between samples) [15]. Also in this case, u 1k does not represent any drug treatment dependence.…”

Section: Drug Treatments Specificitymentioning

confidence: 99%

“…In contrast to the expectation, drug treatments are quite universal. Most of drug treatments (other than (2), (9), (15) and (17)) are separated from control treatments ((2), (9), (15) and (17)) along one direction (red arrow) while diversity among drug treatment spread perpendicular (blue arrow) to the direction only among drug treatments. This suggests that gene expression profiles of genes are altered similarly independent of drug treatment.…”

Section: Drug Treatments Specificitymentioning

confidence: 99%

See 4 more Smart Citations

Universal nature of drug treatment responses in drug-tissue-wide model-animal experiments using tensor decomposition-based unsupervised feature extraction

Taguchi

Turki

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Background: Gene expression profiles of tissues treated with drug recently came to be used for the inference of clinical outcomes. Although it is often successful from the application point of view, gene expression altered by the drug is rarely analyzed in detail because of too many number of genes measured. Method:We apply tensor decomposition (TD) based unsupervised feature extraction (FE) to gene expression profiles of 24 mice tissues treated with 15 drugs.Results: TD based unsupervised FE unexpectedly identified universal feature of 15 drugs, i.e., the effect of 15 drugs are common to some extent. These drugs affect genes in genes-group wide manner, i.e. dependent upon three set of tissue types (Neuronal tissues, muscle tissues, and gastroenterological tissues). For each tissue group, TD based unsupervised FE identified a few tens to a few hundreds genes that are affected by drug treatment. These genes are distinctly expressive between drug treatment and controls as well as between tissues in individual tissue groups and other tissues. Our various enrichment analysis performed guaranteed that the selected genes attributed to individual tissue groups are appreciated.Conclusions: Our TD based unsupervised FE is the promising method to perform integrated analysis of gene expression profiles of multiple tissues treated with multiple drugs in fully unsupervised manner.

show abstract

Section: Td Based Unsupervised Fementioning

confidence: 99%

Section: Td Based Unsupervised Fementioning

confidence: 99%

Section: Td Based Unsupervised Fementioning

confidence: 99%

Section: Drug Treatments Specificitymentioning

confidence: 99%

Section: Drug Treatments Specificitymentioning

confidence: 99%

See 3 more Smart Citations

Universal nature of drug treatment responses in drug-tissue-wide model-animal experiments using tensor decomposition-based unsupervised feature extraction

Taguchi

Turki

2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Suppression of intrahepatic cholangiocarcinoma cell growth by SKI via upregulation of the CDK inhibitor p21

et al. 2022

View full text Add to dashboard Cite

Cholangiocarcinoma (CC) has a poor prognosis and different driver genes depending on the site of onset. Intrahepatic CC is the second-most common liver cancer after hepatocellular carcinoma, and novel therapeutic targets are urgently needed. The present study was conducted to identify novel therapeutic targets by exploring differentially regulated genes in human CC. MicroRNA (miRNA) and mRNA microarrays were performed using tissue and serum samples obtained from 24 surgically resected hepatobiliary tumor cases, including 10 CC cases. We conducted principal component analysis to identify differentially expressed miRNA, leading to the identification of miRNA-3648 as a differentially expressed miRNA. We used an in silico screening approach to identify its target mRNA, the tumor suppressor Sloan Kettering Institute (SKI). SKI protein expression was decreased in human CC cells overexpressing miRNA-3648, endogenous SKI protein expression was decreased in human CC tumor tissues, and endogenous SKI mRNA expression was suppressed in human CC cells characterized by rapid growth. SKI-overexpressing OZ cells (human intrahepatic CC cells) showed upregulation of cyclin-dependent kinase inhibitor p21 mRNA and protein expression and suppressed cell proliferation. Nuclear expression of CDT1 (chromatin licensing and DNA replication factor 1), which is required for the G1/S transition, was suppressed in SKI-overexpressing OZ cells. SKI knockdown resulted in the opposite effects. Transgenic p21-luciferase was activated in SKI-overexpressing OZ cells. These data indicate SKI involvement in p21 transcription and that SKI-p21 signaling causes cell cycle arrest in G1, suppressing intrahepatic CC cell growth. Therefore, SKI may be a potential therapeutic target for intrahepatic CC.

show abstract

Selected Aspects of Interactive Feature Extraction

Grzegorowski¹

2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Unsupervised Feature Extraction Applied to Bioinformatics

Cited by 62 publications

References 0 publications

Universal nature of drug treatment responses in drug-tissue-wide model-animal experiments using tensor decomposition-based unsupervised feature extraction

Universal nature of drug treatment responses in drug-tissue-wide model-animal experiments using tensor decomposition-based unsupervised feature extraction

Suppression of intrahepatic cholangiocarcinoma cell growth by SKI via upregulation of the CDK inhibitor p21

Selected Aspects of Interactive Feature Extraction

Contact Info

Product

Resources

About