Deep neural networks for interpreting RNA-binding protein target preferences

Abstract

Deep learning has become a powerful paradigm to analyze the binding sites of regulatory factors including RNA-binding proteins (RBPs), owing to its strength to learn complex features from possibly multiple sources of raw data. However, the interpretability of these models, which is crucial to improve our understanding of RBP binding preferences and functions, has not yet been investigated in significant detail. We have designed a multitask and multimodal deep neural network for characterizing in vivo RBP targets. The model incorporates not only the sequence but also the region type of the binding sites as input, which helps the model to boost the prediction performance. To interpret the model, we quantified the contribution of the input features to the predictive score of each RBP. Learning across multiple RBPs at once, we are able to avoid experimental biases and to identify the RNA sequence motifs and transcript context patterns that are the most important for the predictions of each individual RBP. Our findings are consistent with known motifs and binding behaviors and can provide new insights about the regulatory functions of RBPs.

Long-read RNA sequencing of human and animal filarial parasites improves gene models and discovers operons

“…We curated a list of 205 genes of interest in B. malayi and D. immitis, consisting of known anthelmintic targets and receptors belonging to druggable protein families in parasitic of reads contained the canonical poly(A) signal (PAS) AAUAA, while A-rich regions were the second most abundant PAS. (C) All PASs occur 10-20 nucleotides upstream of the poly(A) cleavage site, other than AAAAA, which has two A-rich sites at 10-20 and 45-50 nucleotides upstream, and UAAUAA, which may not be a PAS but is a motif for RNA binding proteins [64,65]. (D) All PASs also include an A-rich region at 15 nucleotides upstream of the cleavage site.…”

Graph neural representational learning of RNA secondary structures for predicting RNA-protein interactions

“…Finally, another contribution of our paper is to quantify the impact of and rectify a specific type of sequence bias present in certain CLIP-Seq datasets, originally identified by Kishore et al (2011) and later emphasized in Ghanbari and Ohler (2020) , which artificially inflated the reported prediction accuracy of several approaches published recently.…”

Global Importance Analysis: An Interpretability Method to Quantify Importance of Genomic Features in Deep Neural Networks

“…Recently, deep neural networks (DNNs), predominantly based on convolutional neural networks (CNNs) or convolutionalrecurrent network hybrids, have emerged as a promising alternative, in most cases, improving prediction performance on held-out test data [13][14][15][16][17][18][19] . DNNs are a powerful class of models that can learn a functional mapping between input genomic sequences and experimentally measured labels, requiring minimal feature engineering [20][21][22] .…”

“…At the time, it demonstrated improved performance over PWM-and k-mer-based methods on the 2013-RNAcompete dataset, a standard benchmark dataset that consists of 244 in vitro affinity selection experiments that span across many RBP families 5 . Since then, other deep learning-based methods have emerged, further improving prediction performance on this dataset [23][24][25] and other CLIP-seq-based datasets 11,18,26,27 .…”

“…To validate that DNNs are learning biologically meaningful features, learned representations are visualized and compared to known motifs, previously identified by PWM-and k-mer-based methods 28 . For RBPs, this has been accomplished by visualizing first convolutional layer filters and by attribution methods 13,18,23,24 . First layer filters have been shown to capture motif-like representations, but their efficacy depends highly on choice of model architecture 29 , activation function 30 , and training procedure 31,32 .…”