Unsupervised feature construction and knowledge extraction from genome-wide assays of breast cancer with denoising autoencoders - PubMed (original) (raw)

Unsupervised feature construction and knowledge extraction from genome-wide assays of breast cancer with denoising autoencoders

Jie Tan et al. Pac Symp Biocomput. 2015.

Abstract

Big data bring new opportunities for methods that efficiently summarize and automatically extract knowledge from such compendia. While both supervised learning algorithms and unsupervised clustering algorithms have been successfully applied to biological data, they are either dependent on known biology or limited to discerning the most significant signals in the data. Here we present denoising autoencoders (DAs), which employ a data-defined learning objective independent of known biology, as a method to identify and extract complex patterns from genomic data. We evaluate the performance of DAs by applying them to a large collection of breast cancer gene expression data. Results show that DAs successfully construct features that contain both clinical and molecular information. There are features that represent tumor or normal samples, estrogen receptor (ER) status, and molecular subtypes. Features constructed by the autoencoder generalize to an independent dataset collected using a distinct experimental platform. By integrating data from ENCODE for feature interpretation, we discover a feature representing ER status through association with key transcription factors in breast cancer. We also identify a feature highly predictive of patient survival and it is enriched by FOXM1 signaling pathway. The features constructed by DAs are often bimodally distributed with one peak near zero and another near one, which facilitates discretization. In summary, we demonstrate that DAs effectively extract key biological principles from gene expression data and summarize them into constructed features with convenient properties.

PubMed Disclaimer

Figures

Fig. 1

A) The network structure of denoising autoencoders. B) The distribution of one node's weight vector. C) The distribution of activity values for a node are bimodally distributed. Here we use Node5 as an example.

Fig. 2

Kaplan-Meier plots of disease-specific survival for Node5 (A), ER status (B), Luminal A subtype (C), and Tumor Grade (D) demonstrate that the constructed feature outperforms the other predictors.

Cited by

Wnt/_β_-Catenin, Carbohydrate Metabolism, and PI3K-Akt Signaling Pathway-Related Genes as Potential Cancer Predictors.
Chen P, Shi P, Du G, Zhang Z, Liu L. Chen P, et al. J Healthc Eng. 2019 Oct 20;2019:9724589. doi: 10.1155/2019/9724589. eCollection 2019. J Healthc Eng. 2019. PMID: 31781361 Free PMC article.
Exploring the druggable space around the Fanconi anemia pathway using machine learning and mechanistic models.
Esteban-Medina M, Peña-Chilet M, Loucera C, Dopazo J. Esteban-Medina M, et al. BMC Bioinformatics. 2019 Jul 2;20(1):370. doi: 10.1186/s12859-019-2969-0. BMC Bioinformatics. 2019. PMID: 31266445 Free PMC article.
Accurate Classification of Protein Subcellular Localization from High-Throughput Microscopy Images Using Deep Learning.
Pärnamaa T, Parts L. Pärnamaa T, et al. G3 (Bethesda). 2017 May 5;7(5):1385-1392. doi: 10.1534/g3.116.033654. G3 (Bethesda). 2017. PMID: 28391243 Free PMC article.
Developing and comparing deep learning and machine learning algorithms for osteoporosis risk prediction.
Qiu C, Su K, Luo Z, Tian Q, Zhao L, Wu L, Deng H, Shen H. Qiu C, et al. Front Artif Intell. 2024 Jun 11;7:1355287. doi: 10.3389/frai.2024.1355287. eCollection 2024. Front Artif Intell. 2024. PMID: 38919268 Free PMC article.
Deep Artificial Neural Networks and Neuromorphic Chips for Big Data Analysis: Pharmaceutical and Bioinformatics Applications.
Pastur-Romay LA, Cedrón F, Pazos A, Porto-Pazos AB. Pastur-Romay LA, et al. Int J Mol Sci. 2016 Aug 11;17(8):1313. doi: 10.3390/ijms17081313. Int J Mol Sci. 2016. PMID: 27529225 Free PMC article. Review.

References

1. Wong AK, Park CY, Greene CS, Bongo LA, Guan Y, Troyanskaya OG. Nucleic Acids Research. 2012;40:W484. - PMC - PubMed
1. Schadt EE, Lamb J, Yang X, Zhu J, Edwards S, GuhaThakurta D, Sieberts SK, Monks S, Reitman M, Zhang C, et al. Nature Genetics. 2005;37:710. - PMC - PubMed
1. Sørlie T, Tibshirani R, Parker J, Hastie T, Marron J, Nobel A, Deng S, Johnsen H, Pesich R, Geisler S, et al. Proceedings of the National Academy of Sciences. 2003;100:8418. - PMC - PubMed
1. Vincent P, Larochelle H, Bengio Y, Manzagol P-A. Proceedings of the 25th International Conference on Machine Learning. ACM; New York, NY, USA: 2008. Extracting and composing robust features with denoising autoencoders.
1. Bengio Y. Foundations and trends in Machine Learning. 2009;2:1.

Unsupervised feature construction and knowledge extraction from genome-wide assays of breast cancer with denoising autoencoders - PubMed (original) (raw)