Modeling multi-species RNA modification through multi-task curriculum learning

Abstract

N6-methyladenosine (m6A) is the most pervasive modification in eukaryotic mRNAs. Numerous biological processes are regulated by this critical post-transcriptional mark, such as gene expression, RNA stability, RNA structure and translation. Recently, various experimental techniques and computational methods have been developed to characterize the transcriptome-wide landscapes of m6A modification for understanding its underlying mechanisms and functions in mRNA regulation. However, the experimental techniques are generally costly and time-consuming, while the existing computational models are usually designed only for m6A site prediction in a single-species and have significant limitations in accuracy, interpretability and generalizability. Here, we propose a highly interpretable computational framework, called MASS, based on a multi-task curriculum learning strategy to capture m6A features across multiple species simultaneously. Extensive computational experiments demonstrate the superior performances of MASS when compared to the state-of-the-art prediction methods. Furthermore, the contextual sequence features of m6A captured by MASS can be explained by the known critical binding motifs of the related RNA-binding proteins, which also help elucidate the similarity and difference among m6A features across species. In addition, based on the predicted m6A profiles, we further delineate the relationships between m6A and various properties of gene regulation, including gene expression, RNA stability, translation, RNA structure and histone modification. In summary, MASS may serve as a useful tool for characterizing m6A modification and studying its regulatory code. The source code of MASS can be downloaded from https://github.com/mlcb-thu/MASS.

Geographic encoding of transcripts enabled high-accuracy and isoform-aware deep learning of RNA methylation

“…Geographic encodings were generated according to the latest version of Ensembl transcriptome annotation v104. To deal with imbalanced training data in deep learning models, we up-sampled the positive sites in the training data by 10 times as in previous work ( 41 ). The test data remains unbalanced, so the most informative evaluation metric is the average precision (AP).…”

“…Since experimental approaches for studying RNA modification are expensive and laborious, in silico methods have drawn increasing attention as an alternative avenue and have achieved great success in recent years. To date, more than 100 different approaches ( 22–26 ) have been established for computational prediction of RNA modification sites, including most notably, the iRNA toolkit ( 27–36 ), SRAMP ( 37 ), WHISTLE ( 38 ), Gene2vec ( 39 ), PEA ( 40 ), DeepPromise ( 25 ), MASS ( 41 ), m6Aboost ( 42 ), MultiRM ( 43 ), DeepAc4C ( 44 ), WeakRM ( 45 ), PULSE ( 46 ), NmRF ( 47 ), etc. Among them, the iRNA toolkit ( 27–36 ) developed primarily by Chen, Lin and Chou is the earliest as well as the most versatile toolkit, supporting multiple RNA modification types based on RNA primary sequences and has been widely recognized as the gold standard for benchmarking the accuracy of different RNA modification prediction approaches.…”

DLm6Am: A Deep-Learning-Based Tool for Identifying N6,2′-O-Dimethyladenosine Sites in RNA Sequences

“…Recently, a sensitive and specific approach termed m 6 Am-seq [ 14 ] has been developed to directly profile transcriptome-wide m 6 Am, which can provide high-confidence m 6 Am site data at the single-nucleotide level to develop computational methods of m 6 Am site detection and future functional studies of m 6 Am modification. Moreover, in the field of m 6 A site prediction, MultiRM [ 17 ], DeepM6ASeq [ 20 ], and MASS [ 21 ] were successfully developed using an effective hybrid framework embedding with CNN and LSTM and achieved promising performance. Inspired by these single-nucleotide m 6 Am data and successful applications of deep learning frameworks, here, we present DLm6Am, an attention-based ensemble deep-learning framework to accurately identify m 6 Am sites.…”

m6A-TSHub: unveiling the context-specific m6A methylation and m6A-affecting mutations in 23 human tissues

“…To date, more than 120 computational approaches have been developed for the computational identification of RNA modifications [21, 22] from the primary RNA sequences. These include the iRNA toolkits [23–31], MultiRM [32], DeepPromise [22], RNAm5CPred [33], SRAMP [11], Gene2vec [34], PEA [35], PPUS [36], WHISTLE [37], m5UPred [38], WeakRM frameworks [39, 40], m6ABoost [41], PULSE [42], m6AmPred [43], BERMP [44] and MASS [45]. Together, these efforts greatly advanced our understanding of multiple RNA modifications at different RNA regions and in various species (see recent reviews [22, 46–48]).…”

“…MultiRM [32], DeepPromise [22], RNAm5CPred [33], SRAMP [11], Gene2vec [34], PEA [35], PPUS [36], WHISTLE [37], m5UPred [38], WeakRM frameworks [39,40], m6ABoost [41], PULSE [42], m6AmPred [43], BERMP [44] and MASS [45]. Together, these efforts greatly advanced our understanding of multiple RNA modifications at different RNA regions and in various species (see recent reviews [22,[46][47][48]).…”