Research ArticlePrognostic prediction of carcinoma by a differential-regulatory-network-embedded deep neural network
Section snippets
Background
Carcinoma is one of the common diseases that cause death in humans. Cancer occurrence and development is usually a multi-factor, multi-step complex process (Hanahan and Weinberg, 2011; Greaves and Maley, 2012). According to the latest global cancer statistics in 2018, there are an estimated 18.19 million new cancer cases and 9.6 million cancer deaths worldwide (Bray et al., 2018). Therefore, accurate prediction of the survival time is essential for precise cancer diagnosis and treatment, which
Data sets
Three public HCC data sets from GSE10143 (Yang et al., 2013), GSE14520 (Hoshida et al., 2008) and TCGA (Roessler et al., 2012) are used in our study as shown as Table 1 shows each set of data consists of a gene expression profile and contain clinical data for disease state measurements.
GSE10143 has 80 cancer patient tissue samples and 82 normal tissue samples, containing 6100 characteristic genes. GSE14520 has 221 cancer patients tissue samples and 210 normal tissue samples, containing 13,050
DRE-DNN training results
We apply the proposed DRE-DNN and norml DNN model to these three datasets from GEO and TCGA databases.
In the training set, we performed 10 verification experiments. Area Under the Curve (AUC) value can objectively reflect the ability of the model to comprehensively predict positive and negative samples, and consider the impact of unbalanced data. The greater the AUC value, the stronger the ability to accurately predict. Table 2 shows that the average AUC values for prediction results of the
Conclusions
Deep learning method is very promising and non-trivial in bioinformatics studies. Embedding valuable genetic information into DNN to predict prognosis of cancer helps to prevent the problem of overfitting when dealing with the high-dimensional transcriptomic data. DRE-DNN integrates differential regulatory analysis based on gene co-expression network and DNN method, which makes a better survival prediction for each cancer data set.
In this work, we embed regulatory information into DNN and
Additional files
All additional files are available at: https://github.com/biohitszcs2019/DREDNN
Authors’ contributions
JL and YP designed the study, performed bioinformatics analysis and drafted the manuscript. All of the authors performed the analysis and participated in the revision of the manuscript. JL and YW conceived of the study, participated in its design and coordination and drafted the manuscript. All authors read and approved the final manuscript.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work was supported by the grants from the National “863” Key Basic Research Development Program (2014AA021505), the National Key Research Program (2017YFC1201201) and the startup grant of Harbin Institute of Technology (Shenzhen).
References (33)
- et al.
Deep convolutional neural networks for breast cancer screening
Comput. Methods Programs Biomed.
(2018) - et al.
Hallmarks of cancer: the next generation
Cell
(2011) - et al.
A novel integrated gene coexpression analysis approach reveals a prognostic three-transcription-factor signature for glioma molecular subtypes
BMC Syst. Biol.
(2016) - et al.
The Eighth Edition AJCC cancer staging manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging
CA Cancer J. Clin.
(2017) - et al.
Deep learning for computational biology
Mol. Syst. Biol.
(2016) - et al.
Adalsteinsson e: fast image reconstruction with L2-regularization
J. Magn. Reson. Imaging
(2014) - et al.
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
CA-Cancer J. Clin.
(2018) - et al.
Deep learning-based multi-omics integration robustly predicts survival in liver cancer
Clin. Cancer Res.
(2018) - et al.
Protein networks as logic functions in development and cancer
PLoS Comput. Biol.
(2011) Survival curves
Rev Chir Orthop
(1990)
Clonal evolution in cancer
Nature
The coming era of artificial intelligence in biological data science
BMC Bioinformatics
Gene expression in fixed tissues and outcome in hepatocellular carcinoma
New England J. Med. Surg. Collat. Branches Sci.
Package ‘survminer’
Network-based penalized regression with application to genomic data
Biometrics
Adam: a method for stochastic optimization
Comput. Sci.
Cited by (9)
Residual useful life prognosis of equipment based on modified hidden semi-Markov model with a co-evolutional optimization method
2023, Computers and Industrial EngineeringIntersection Three Feature Selection and Machine Learning Approaches for Cancer Classification
2023, Proceedings of 2023 International Conference on System Science and Engineering, ICSSE 2023Deep learning-based microarray cancer classification and ensemble gene selection approach
2022, IET Systems Biology
- 1
These authors contributed equally to this work.