High Class-Imbalance in pre-miRNA Prediction: A Novel Approach Based on deepSOM

Abstract

The computational prediction of novel microRNA within a full genome involves identifying sequences having the highest chance of being a miRNA precursor (pre-miRNA). These sequences are usually named candidates to miRNA. The well-known pre-miRNAs are usually only a few in comparison to the hundreds of thousands of potential candidates to miRNA that have to be analyzed, which makes this task a high class-imbalance classification problem. The classical way of approaching it has been training a binary classifier in a supervised manner, using well-known pre-miRNAs as positive class and artificially defining the negative class. However, although the selection of positive labeled examples is straightforward, it is very difficult to build a set of negative examples in order to obtain a good set of training samples for a supervised method. In this work, we propose a novel and effective way of approaching this problem using machine learning, without the definition of negative examples. The proposal is based on clustering unlabeled sequences of a genome together with well-known miRNA precursors for the organism under study, which allows for the quick identification of the best candidates to miRNA as those sequences clustered with known precursors. Furthermore, we propose a deep model to overcome the problem of having very few positive class labels. They are always maintained in the deep levels as positive class while less likely pre-miRNA sequences are filtered level after level. Our approach has been compared with other methods for pre-miRNAs prediction in several species, showing effective predictivity of novel miRNAs. Additionally, we will show that our approach has a lower training time and allows for a better graphical navegability and interpretation of the results. A web-demo interface to try deepSOM is available at http://fich.unl.edu.ar/sinc/web-demo/deepsom/.

microRNA analysis of Taenia crassiceps cysticerci under praziquantel treatment and genome-wide identification of Taenia solium miRNAs

“…Additionally, we used an independent approach based on a deep architecture of self-organizing maps (SOMs), called miRNA-SOM (Kamenetzky et al, 2016), in order to confirm miRNA predictions and identify further miRNAs. Sequences with a minimum free energy threshold of -20 kcal/mol and single-loop folded sequences were selected according to the miRNA biogenesis model (Bartel, 2004) and then the best candidates were sequentially filtered in the SOM layers (Kamenetzky et al, 2016;Stegmayer et al, 2016).…”

PESM: predicting the essentiality of miRNAs based on gradient boosting machines and sequences

“…We use the module RNAlib of Vienna RNA Package to intrinsic folding quantitative measures P(S), nP(S), Q(s), nQ(s), D(s) and nD(s) [62]. These structure features and Vienna RNA Package have been broadly used in both miRNA prediction and pre-miRNA prediction [63][64][65]. As a result, our method consists of 38 features.…”

Improved Pre-miRNAs Identification Through Mutual Information of Pre-miRNA Sequences and Structures

“…The ML-based methods usually model pre-miRNA identification as a binary classification problem to discriminate real and pseudo-pre-miRNAs. Widely used ML-based algorithms include support vector machines (SVMs) (Xue et al, 2005; Helvik et al, 2007; Huang et al, 2007; Wang Y. et al, 2011; Lei and Sun, 2014; Lopes et al, 2014; Wei et al, 2014; Liu et al, 2015b; Khan et al, 2017), back-propagation and self-organizing map (SOM) neural networks (Stegmayer et al, 2016; Zhao et al, 2017), linear genetic programming (Markus and Carsten, 2007), hidden Markov model (Agarwal et al, 2010), random forest (RF) (Jiang et al, 2007; Kandaswamy et al, 2011; Lin et al, 2011), covariant discrimination (Chou and Shen, 2007; Lopes et al, 2014), Naive Bayes (Lopes et al, 2014), and deep learning (Mathelier and Carbone, 2010). For example, Yousef et al (2006) Peng et al (2018) used a Bayesian classifier for pre-miRNA recognition, which has demonstrated effectiveness in recognizing pre-miRNAs in the genomes of different species.…”