Fast and accurate microRNA search using CNN

Abstract

BackgroundThere are many different types of microRNAs (miRNAs) and elucidating their functions is still under intensive research. A fundamental step in functional annotation of a new miRNA is to classify it into characterized miRNA families, such as those in Rfam and miRBase. With the accumulation of annotated miRNAs, it becomes possible to use deep learning-based models to classify different types of miRNAs. In this work, we investigate several key issues associated with successful application of deep learning models for miRNA classification. First, as secondary structure conservation is a prominent feature for noncoding RNAs including miRNAs, we examine whether secondary structure-based encoding improves classification accuracy. Second, as there are many more non-miRNA sequences than miRNAs, instead of assigning a negative class for all non-miRNA sequences, we test whether using softmax output can distinguish in-distribution and out-of-distribution samples. Finally, we investigate whether deep learning models can correctly classify sequences from small miRNA families.ResultsWe present our trained convolutional neural network (CNN) models for classifying miRNAs using different types of feature learning and encoding methods. In the first method, we explicitly encode the predicted secondary structure in a matrix. In the second method, we use only the primary sequence information and one-hot encoding matrix. In addition, in order to reject sequences that should not be classified into targeted miRNA families, we use a threshold derived from softmax layer to exclude out-of-distribution sequences, which is an important feature to make this model useful for real transcriptomic data. The comparison with the state-of-the-art ncRNA classification tools such as Infernal shows that our method can achieve comparable sensitivity and accuracy while being significantly faster.ConclusionAutomatic feature learning in CNN can lead to better classification accuracy and sensitivity for miRNA classification and annotation. The trained models and also associated codes are freely available at https://github.com/HubertTang/DeepMir.

Heuristic hyperparameter optimization of deep learning models for genomic prediction

“…In genomic data, linkage disequilibrium is a nonrandom relationship of alleles at different physical locations, which is a sensitive indicator that structures a genome (Slatkin 2008). Also, Tang and Sun (2019) argued that CNN could be utilized to extract motifs from homologous sequences, where motifs are essential features for distinguishing different sequence families. Given a dataset with spatial structure, CNN potentially has advantage over MLP that CNN can deal with local connectivity.…”

A framework for predicting lncRNAs expression in human dendritic cells in response to M. tuberculosis infection

“…Therefore, as shown in Figure 1, CNNs technique is proposed to be implemented in identifying and classifying between lncRNAs and mRNAs being expressed in human DCs. We intend to explore the capability of CNNs to extract information from one-dimensional biological sequences data as discussed by [36], [37].…”

Sentences, Words, Attention: A “Transforming” Aphorism for the Discovery of pre-miRNA Regions across Plant Genomes

“…miRNA biology has very recently witnessed some of them for pre-miRNA discovery, which though not developed exclusively for plants, but can be trained on the plant specific data also. This includes appraoches involving Boltzman machines based deep learning DP-miRNA/deepBN (27), deep learning based self organizing maps (SOM) (28), convolution neural nets (CNN) based miRNA classifier like deepMiR (29), convolutional deep residual networks (30), long-short term memory (LSTM) based pre-miRNA classifier like deepMiRGene (31).…”