BP-GAN: Interpretable Human Branchpoint Prediction Using Attentive Generative Adversarial Networks

Abstract

Branchpoints (BPs) are essential sequence elements of ribonucleic acids (RNAs) in splicing, which is the process of creating a messenger RNA (mRNA) that is translated into proteins. This study proposes to develop deep neural networks for BP prediction. Extensive previous studies have shown that the existence of BP sites depends on sequence patterns called motifs; hence, the prediction model must accurately explain its decisions in terms of motifs. Existing approaches utilized either handcrafted features for interpretable, though less accurate, predictions or deep neural networks that were accurate but difficult to explain. To address the aforementioned difficulties, the proposed method incorporates 1) generative adversarial networks (GANs) to learn the latent structure of RNA sequences, and 2) an attention mechanism to learn sequence-positional long-term dependency for accurate prediction and interpretation. Our method achieves highly satisfying results in various performance metrics with adequate interpretability. We demonstrated that, without any prior biological knowledge, BP prediction by the proposed method is closely related to three motifs, the consensus sequence surrounding BPs, polypyrimidine tract, and 3' splice site, that are well-established in molecular biology. INDEX TERMS Branchpoint prediction, deep neural networks, generative adversarial networks, interpretability.

Transformer Architecture and Attention Mechanisms in Genome Data Analysis: A Comprehensive Review

“…Several studies have also focused on the prediction of specific types of phenotypes. For instance, Lee et al [79] proposed BP-GAN, a model that uses generative adversarial networks (GANs) combined with an attention mechanism for predicting RNA Branchpoints (BPs). These studies have shown the potential of deep learning models in predicting specific types of phenotypes.…”

“…Specifically, attention-based models have been employed to predict gene expression levels using histone modification data, such as Chromoformer [70], TransferChrome [71], and a hybrid convolutional and bi-directional LSTM network [73]. Additionally, researchers have explored the prediction of specific phenotypes, such as toehold switch functions [74] and RNA Branchpoints [79], showcasing the versatility and potential of deep learning with attention mechanisms in gene expression and phenotype prediction.…”

Recent Advances in Generative Adversarial Networks for Gene Expression Data: A Comprehensive Review

“…The BP-GAN [60] stands out in predicting branchpoints in RNA sequences due to its ability to capture the latent structure of RNA sequences and the sequence-positional long-term dependency. The generator in BP-GAN learns to generate synthetic RNA sequences that reflect the intrinsic latent structure and long-term dependencies, while the discriminator ensures the quality of these synthetic sequences by distinguishing them from real sequences.…”

Overall Dynamic Optimization of Metal Mine Technical Indicators Considering Spatial Distribution of Ore Grade Using AADE

“…A BP neural network [34][35][36][37] was used to determine the relationships between copper mining grade and concentrate grade and between concentrate ratio and concentrate grade in the Yinshan copper mine. A total of 630 data items were collected; the training sample consisted of the first 530 data items, and the test sample was the remaining 100 items; mining grade and concentrate ratio were used as inputs, and the copper ore concentrate grade was the output.…”