Abstract
Computational biology plays a significant role in the discovery of new biomarkers, the analyses of disease states and the validation of potential biomarkers. Biomarkers are used to measure the progress of disease or the physiological effects of therapeutic intervention in the treatment of disease. They are also used as early warning signs for various diseases such as cancer and inflammatory diseases. In this review, we outline recent progresses of computational biology application in research on biomarkers discovery. A brief discussion of some necessary preliminaries on machine learning techniques (e.g., clustering and support vector machines—SVM) which are commonly used in many applications to biomarkers discovery is given and followed by a description of biological background on biomarkers. We further examine the integration of computational biology approaches and biomarkers. Finally, we conclude with a discussion of key challenges for computational biology to biomarkers discovery.
Highlights
Machine learning is the subfield of artificial intelligence which focuses on methods to construct computer programs that learn from experience with respect to some class of tasks and a performance measure [1]
To the remaining genes the gene expression network analysis tool (GXNA) [48] is applied to form n clusters of genes that are highly connected in the network
Yousef et al [39] compare the performance of Support vector machines (SVMs)-RCE against the popular SVM-RFE method to reported in most cases that SVM-RCE is with better results as in average of 6 datasets is 96% while SVM-RFE with 92% accuracy
Summary
Machine learning is the subfield of artificial intelligence which focuses on methods to construct computer programs that learn from experience with respect to some class of tasks and a performance measure [1]. Clustering is a simple classical method of the unsupervised learning, which partitions the data set into clusters, so that the data in each subset share some common trait according to some defined distance measure. Most of the unsupervised learning methods use a measure of similarity between patterns in order to group them into clusters. The simplest of these involves defining a distance between patterns. Given a training set of labeled examples ( xi , yi ) i = 1, ,l where xi ∈ R t and yi ∈{+1, −1} , the support vector machines (SVMs) find the separating hyper-plane of the form w ⋅ x + b= 0 w ∈ Rt ,b ∈ R. One could use the following formula as a predictor for a new instance: f= ( x) sign (w⋅ x + b) (for more information see Vapnik [15])
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
