IRSOM2: a web server for predicting bifunctional RNAs

Abstract

Recent advances have shown that some biologically active non-coding RNAs (ncRNAs) are actually translated into polypeptides that have a physiological function as well. This paradigm shift requires adapted computational methods to predict this new class of 'bifunctional RNAs'. Previously, we developed IRSOM, an open-source algorithm to classify non-coding and coding RNAs. Here, we use the binary statistical model of IRSOM as a ternary classifier, called IRSOM2, to identify bifunctional RNAs as a rejection of the two other classes. We present its easy-to-use web interface, which allows users to perform predictions on large datasets of RNA sequences in a short time, to re-train the model with their own data, and to visualize and analyze the classification results thanks to the implementation of self-organizing maps (SOM). We also propose a new benchmark of experimentally validated RNAs that play both protein-coding and non-coding roles, in different organisms. Thus, IRSOM2 showed promising performance in detecting these bifunctional transcripts among ncRNAs of different types, such as circRNAs and lncRNAs (in particular those of shorter lengths). The web server is freely available on the EvryRNA platform: https://evryrna.ibisc.univ-evry.fr.

Graphical Abstract

Figure 1.

Graphical output of the IRSOM2 web server. (A) Representation of the resulting self-organizing map, as a grid of 10 × 10 neurons. (B) For each input sequence, the table gives the map coordinates (x,y) of the cluster (neuron) to which it belongs (the Best Matching Unit, ‘BMU’ column) along with its predicted class and the two associated probabilities. (C) Bar plots that display how the probabilities of being coding or noncoding are distributed among the three RNA categories. (D, E) Optional plots of the feature profiles: ORF and codon bias, respectively. The coordinates of the 10 × 10 box plots correspond to those of the neurons in the grid.

Figure 2.

Performance of IRSOM2 on the experimentally validated SPENCER dataset of cancer-specific bifunctional RNAs (n = 733). (A) Pie chart representing the predictions of IRSOM2 on the bifunctional RNAs from SPENCER. (B) True positive rate (TPR) depending on the sequence length. The TPR is represented in blue. The green surface represents the cumulative distribution of sequence lengths. The red line represents the 33.3% TPR of a random ternary classifier. (C, D) ‘Zoomed views’ representing the subsets of sequences shorter than 5000 nt (n = 707) and 500 nt (n = 119), respectively. (E) Barplot representing the performance of IRSOM2 on different subsets of SPENCER dataset, which correspond to different ranges of RNA sequence lengths (values in nt). The term ‘macroRNAs’ designates lncRNAs longer than 10 kb. The size of each subset is given below the intervals.