A Review on Recent Computational Methods for Predicting Noncoding RNAs

Table 6

Methods to predict lncRNAs.

Name	Feature	Prediction algorithm
Estimating lincRNome size for human [63]	lincRNA numbers validated experimentally in human and mouse, and their overlap lincRNA number	System of nonlinear equations
Classifying human lncRNA [64]	RNA sequence-structure patterns (RSSPs) describing 42 highly structured families, motif binding sites extracted as 1314 Position-Weight Matrices (PWMs), all -words of length , the sequence complexity	Classifying human lncRNA by being able (or disable) to bind the polycomb repressive complex (PRC2), SVM with linear kernel
Identify, classify, and localize maize lncRNAs [65]	Transcript length, open reading frame (ORF) size, and homology with known proteins	SVM
The GENCODE v7 catalog of human lncRNA [66]	Lack of homology with known proteins, no reasonable-sized open reading frame (ORF), and no high conservation, confirmed by PhyloCSF through the majority of exons conserved promoters	Manual annotation and pattern recognition
Highly conserved large noncoding RNAs [67]	Chromatin signatures “K4–K36” domain	Maximum CSF score observed across the entire genomic locus