Lung cancer gene expression database analysis incorporating prior knowledge with support vector machine-based classification method

Our plain-English summary of Lung cancer gene expression database analysis incorporating prior knowledge with support vector machine-based classification method, by P. Guan, D. Huang, M. He, et al (Journal of Experimental & Clinical Cancer Research, 2009).

The study

• Authors: P. Guan, D. Huang, M. He, B. Zhou
• Journal: Journal of Experimental & Clinical Cancer Research
• Published: 2009;28(1):103
• Cited by: 20 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: A reliable and precise classification is essential for successful diagnosis and treatment of cancer. Gene expression microarrays have provided the high-throughput platform to discover genomic biomarkers for cancer diagnosis and prognosis. Rational use of the available bioinformation can not only effectively remove or suppress noise in gene chips, but also avoid one-sided results of separate experiment. However, only some studies have been aware of the importance of prior information in cancer classification.

Methods: Together with the application of support vector machine as the discriminant approach, we proposed one modified method that incorporated prior knowledge into cancer classification based on gene expression data to improve accuracy. A public well-known dataset, Malignant pleural mesothelioma and lung adenocarcinoma gene expression database, was used in this study. Prior knowledge is viewed here as a means of directing the classifier using known lung adenocarcinoma related genes. The procedures were performed by software R 2.80.

Results: The modified method performed better after incorporating prior knowledge. Accuracy of the modified method improved from 98.86% to 100% in training set and from 98.51% to 99.06% in test set. The standard deviations of the modified method decreased from 0.26% to 0 in training set and from 3.04% to 2.10% in test set.

Conclusion: The method that incorporates prior knowledge into discriminant analysis could effectively improve the capacity and reduce the impact of noise. This idea may have good future not only in practice but also in methodology.

In plain English

This was a statistical methods study, not a study of asbestos exposure. The authors tested a computer method for telling malignant pleural mesothelioma apart from lung adenocarcinoma using gene activity measurements. They reanalyzed a public dataset of 12 533 genes measured in 31 malignant pleural mesothelioma samples and 150 lung adenocarcinoma samples. A gene-selection method called PAM picked between 4 and 12 genes in each run. The authors also searched articles published in the journal Cancer Research from January 2000 to March 2009. They found 23 genes reported as related to lung adenocarcinoma, and 8 of these passed a statistical test. They added these 'prior knowledge' genes to a support vector machine classifier. With the added genes, accuracy rose from 98.86% to 100% in the training set and from 98.51% to 99.06% in the test set. The authors reported these differences as statistically significant. They concluded that adding prior knowledge could improve classification and reduce the effect of noise. They noted that how to choose and incorporate prior knowledge, and the minimum number of genes needed, were unresolved.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Computational methods study: reanalysis of a public gene expression microarray dataset, comparing a support vector machine classifier with and without genes taken from published literature (prior knowledge)
• Who was studied: A public Affymetrix microarray dataset (Gordon et al.) of 12 533 genes' expression profiles from 31 malignant pleural mesothelioma (MPM) and 150 lung adenocarcinoma (ADCA) samples. The original training set was 16 ADCA and 16 MPM samples.
• Exposure: No asbestos or other exposure was studied. The comparison was between classification methods: genes selected by PAM alone versus PAM genes combined with lung adenocarcinoma-related genes taken from articles in Cancer Research (Jan 2000 to March 2009).
• Main results:
  • The number of genes selected by PAM ranged from 4 to 12 (average 7.81, standard deviation 2.21). CEACAM6, calretinin, VAC-β and TACSTD1 were always selected.
  • Twenty-three lung adenocarcinoma-related genes were found in the literature, and eight passed the multiple testing procedure in the training set.
  • With prior knowledge, classification accuracy improved from 98.86% to 100% in the training set and from 98.51% to 99.06% in the test set.
  • Standard deviation of accuracy decreased from 0.26% to 0 in the training set and from 3.04% to 2.10% in the test set.
  • When a correlated copy of VAC-β was added to simulate overlap (complete, strong, and minor correlation), accuracy was 99.12%, 99.28% and 99.23% (standard deviations 2.04%, 2.04%, 1.93%). Figure 3 shows these as test-set values, with 100% in the training set.
  • McNemar's test comparing 'no prior knowledge' with each of the four prior-knowledge situations, in both training and test sets, found all differences statistically significant.
• The authors' conclusion: The authors concluded that incorporating prior knowledge into discriminant analysis could improve classification capacity and reduce the impact of noise.
• Limitations the authors state:
  • How to incorporate prior biological knowledge, and where to obtain it, was a main limitation.
  • The minimum number of predictor genes is not known.

Topics

The National Library of Medicine indexes this study under subjects our site covers: Lung cancer.

Medical terms in this study, explained in our Lexicon: Lung Cancer, Biomarker, Prognosis, Pleural Mesothelioma, Lung Adenocarcinoma.

Related studies

• Gene expression profiling and gene copy-number changes in malignant mesothelioma cell lines — C. Zanazzi, R. Hersmus, et al. (2007)
• Gene expression profile of aquaporin 1 and associated interactors in malignant pleural mesothelioma — R. Jagirdar, E.I. Solenov, et al. (2013)
• Identification of intelectin overexpression in malignant pleural mesothelioma by serial analysis of gene expression (SAGE) — A. Wali, P.J. Morin, et al. (2005)
• A multi-class predictor based on a probabilistic model: application to gene expression profiling-based diagnosis of thyroid tumors — N. Yukinawa, S. Oba, et al. (2006)
• Global gene expression profiling of human pleural mesotheliomas: identification of matrix metalloproteinase 14 (MMP-14) as potential tumour target — S. Crispi, R.A. Calogero, et al. (2009)
• Gene expression profiles reveal increased mClca3 (Gob5) expression and mucin production in a murine model of asbestos-induced fibrogenesis — T. Sabo-Attwood, M. Ramos-Nino, et al. (2005)

Browse more research in Journal of Experimental & Clinical Cancer Research, or return to medical research.