A common set of distinct features that characterize noncoding RNAs across multiple species

Abstract

To find signature features shared by various ncRNA sub-types and characterize novel ncRNAs, we have developed a method, RNAfeature, to investigate >600 sets of genomic and epigenomic data with various evolutionary and biophysical scores. RNAfeature utilizes a fine-tuned intra-species wrapper algorithm that is followed by a novel feature selection strategy across species. It considers long distance effect of certain features (e.g. histone modification at the promoter region). We finally narrow down on 10 informative features (including sequences, structures, expression profiles and epigenetic signals). These features are complementary to each other and as a whole can accurately distinguish canonical ncRNAs from CDSs and UTRs (accuracies: >92% in human, mouse, worm and fly). Moreover, the feature pattern is conserved across multiple species. For instance, the supervised 10-feature model derived from animal species can predict ncRNAs in Arabidopsis (accuracy: 82%). Subsequently, we integrate the 10 features to define a set of noncoding potential scores, which can identify, evaluate and characterize novel noncoding RNAs. The score covers all transcribed regions (including unconserved ncRNAs), without requiring assembly of the full-length transcripts. Importantly, the noncoding potential allows us to identify and characterize potential functional domains with feature patterns similar to canonical ncRNAs (e.g. tRNA, snRNA, miRNA, etc) on ∼70% of human long ncRNAs (lncRNAs).

Critical roles of long noncoding RNAs in Drosophila spermatogenesis

“…To this end, we first analyzed published gene expression data and identified 120 lncRNAs specifically and/or highly expressed in fly testis . To comprehensively identify all lncRNAs expressed in fly, we used a previously developed computational algorithm Gerstein et al 2014;Hu et al 2015) to predict 121 novel intergenic lncRNAs with no overlap with protein-coding genes; the predictions were based on RNA-seq data and RNA structure information ( Fig. 1A; Supplemental Fig.…”

“…We first collected the annotated ncRNAs from FlyBase r5.45 and then adopted a machine learning method to predict novel lncRNAs in Drosophila Gerstein et al 2014;Hu et al 2015). In this method, we used multiple features (e.g., sequence, structure, and expression data) to train a random forest model.…”

“…In this method, we used multiple features (e.g., sequence, structure, and expression data) to train a random forest model. The model used the lncRNAs annotated in FlyBase as the training set, and then made predictions throughout the whole genome (Hu et al 2015). Subsequently, the annotated (known) and predicted novel lncRNAs were filtered and classified based on their genomic locations.…”

COME: a robust coding potential calculation tool for lncRNA identification and characterization based on multiple features

“…Based on the composed matrix, we applied a balanced random forest (BRF) algorithm (7,27) to train on the annotated coding (mRNAs) and non-coding transcripts (lncRNAs). The predicted probability of being a coding transcript was defined as COME's coding potential score for each given transcript.…”

“…Moreover, ribosome profiling data suggest that ribosomes may have different binding patterns on mRNAs and lncRNAs (26). Our previous study demonstrated that integration of experiment-based and sequence-derived features could enable classification of ncRNAs from coding sequences with high accuracy (27). However, the model focused only on canonical ncRNAs, and thus, can only predict the local regions of ∼70% of human lncRNAs.…”