Feature Selection for MicroRNA Target Prediction - Comparison of One-Class Feature Selection Methodologies

Abstract

Abstract:Traditionally, machine learning algorithms build classification models from positive and negative examples. Recently, one-class classification (OCC) receives increasing attention in machine learning for problems where the negative class cannot be defined unambiguously. This is specifically problematic in bioinformatics since for some important biological problems the target class (positive class) is easy to obtain while the negative one cannot be measured. Artificially generating the negative class data can be based on unreliable assumptions. Several studies have applied two-class machine learning to predict microRNAs (miRNAs) and their target. Different approaches for the generation of an artificial negative class have been applied, but may lead to a biased performance estimate. Feature selection has been well studied for the two-class classification problem, while fewer methods are available for feature selection in respect to OCC. In this study, we present a feature selection approach for applying one-class classification to the prediction of miRNA targets. A comparison between one-class and two-class approaches is presented to highlight that their performance are similar while one-class classification is not based on questionable artificial data for training and performance evaluation. We further show that the feature selection method we tried works to a degree, but needs improvement in the future. Perhaps it could be combined with other approaches.

Delineating the impact of machine learning elements in pre-microRNA detection

“…OneClass performed worst which is likely due to feature selection which has a strong impact on OCC (Yousef et al, 2016; Yousef, Allmer & Khalifa, 2016). While the median of OCC represents random guessing, RF is only 0.05 points away from full accuracy.…”

Species Categorization via MicroRNAs - Based on 3’UTR Target Sites using Sequence Features

“…K-mers are short stretches of nucleotides of length k (also termed n-grams or words). Such sequencebased features were used for ab initio pre-miRNA detection, before, and may also be useful for target prediction (Yousef, Allmer and Khalifa, 2016b). Formally, a 1-mer is one element of the relevant alphabet, here {A, U, C, G}.…”

“…There is no dataset holding the guarantee not to contain target sites for miRNAs which confounds their computational prediction (Hamzeiy, Allmer and Yousef, 2014). A viable approach to remove the dependency on negative data is to use one-class classification (Yousef, Allmer and Khalifa, 2016b). For the computational detection of miRNA targets (Peterson et al, 2014), generally the miRNA:mRNA duplex is considered.…”

Peer Review #2 of "The impact of feature selection on one and two-class classification performance for plant microRNAs (v0.1)"

“…Moreover, the pseudo negative sequences were injected as unknown class during testing. We employed k-means in this study as previously described [43] since it performed well in respect to OCC although it is a clustering algorithm. During learning, labeled examples are clustered (miRNAs and unknown) and during testing and in prediction, the label of the closest cluster is assigned to the sample.…”