G4-Attention: Deep Learning Model with Attention for predicting DNA G-Quadruplexes

Abstract

G-quadruplexes (G4s) are the four-stranded non-canonical nucleic acid secondary structures, formed by the stacking arrangement of the guanine tetramers. They are involved in a wide range of biological roles because of their exceptionally unique and distinct structural characteristics. After the completion of the human genome sequencing project, a lot of bio-informatics algorithms were introduced to predict the active G4s regions in vitro based on the canonical G4 sequence elements, G-richness, and G-skewness, as well as the non-canonical sequence features. Recently, sequencing techniques like G4-seq and G4-ChIP-seq were developed to map the G4s in vitro, and in vivo respectively at a few hundred base resolution. Subsequently, several machine learning and deep learning approaches were developed for predicting the G4 regions using the existing databases. However, their prediction models were simplistic, and the prediction accuracy was notably poor. In response, here, we propose a novel convolutional neural network with Bi-LSTM and attention layers, named G4-Attention, to predict the G4 forming sequences with improved accuracy. G4-Attention achieves high accuracy and attains state-of-the-art results in the G4 propensity and mismatch score prediction task in comparison to other available benchmark models in the literature. Besides the balanced dataset, the developed model can predict the G4 regions accurately in the highly class-imbalanced datasets. Furthermore, the model achieves a significant improvement in the cell-type-specific G4 prediction task. In addition, G4-Attention trained on the human genome dataset can be applied to any non-human genomic DNA sequences to predict the G4 formation propensities accurately. We have also added interpretability analysis of our model to gain further insights. Author summaryG-quadruplex, a non-canonical secondary nucleic acid structure, has emerged as a potential pharmacological target because of its significant implication in several human diseases including cancer, aging, neurological disorders, etc. Despite numerous computational algorithm developments, the prediction of G4 regions accurately in different organisms including humans still remains a challenging task. To address this, in this work, we have presented a novel advanced deep learning architecture called G4-Attention for predicting DNA G-quadruplexes in different organisms including humans. To the best of our knowledge, we are the first to incorporate Bi-LSTM and attention layers on top of a CNN architecture in a deep learning model (G4-Attention) for predicting G4-forming sequences. Our developed model outperforms existing algorithms and achieves current state-of-the-art (SOTA) results in G4 propensity and mismatch score prediction tasks. In addition, the developed model achieves superior results across non-human genomes, class-imbalanced datasets, and cell line-specific datasets. Lastly, G4-Attention can identify key features for understanding the G4 formation mechanism.