Imbalance deep multi‐instance learning for predicting isoform–isoform interactions

LMGATCDA: Graph Neural Network With Labeling Trick for Predicting circRNA-Disease Associations

Previous studies have proven that circular RNAs (circRNAs) are inextricably connected to the etiology and pathophysiology of complicated diseases. Since conventional biological research are frequently small-scale, expensive, and time-consuming, it is essential to establish an efficient and reasonable computation-based method to identify disease-related circRNAs. In this article, we proposed a novel ensemble model for predicting probable circRNA-disease associations based on multi-source similarity information(LMGATCDA). In particular, LMGATCDA first incorporates information on circRNA functional similarity, disease semantic similarity, and the Gaussian interaction profile (GIP) kernel similarity as explicit features, along with node-labeling of the three-hop subgraphs extracted from each linked target node as graph structural features. After that, the fused features are used as input, and further implied features are extracted by graph sampling aggregation (GraphSAGE) and multi-hop attention graph neural network (MAGNA). Finally, the prediction scores are obtained through a fully connected layer. With five-fold cross-validation, LMGATCDA demonstrated excellent competitiveness against gold standard data, reaching 95.37% accuracy and 91.31% recall with an AUC of 94.25% on the circR2Disease benchmark dataset. Collectively, the noteworthy findings from these case studies support our conclusion that the LMGATCDA model can provide reliable circRNA-disease associations for clinical research while helping to mitigate experimental uncertainties in wet-lab investigations.

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

The emergence and rapid development of deep learning, specifically transformer-based architectures and attention mechanisms, have had transformative implications across several domains, including bioinformatics and genome data analysis. The analogous nature of genome sequences to language texts has enabled the application of techniques that have exhibited success in fields ranging from natural language processing to genomic data. This review provides a comprehensive analysis of the most recent advancements in the application of transformer architectures and attention mechanisms to genome and transcriptome data. The focus of this review is on the critical evaluation of these techniques, discussing their advantages and limitations in the context of genome data analysis. With the swift pace of development in deep learning methodologies, it becomes vital to continually assess and reflect on the current standing and future direction of the research. Therefore, this review aims to serve as a timely resource for both seasoned researchers and newcomers, offering a panoramic view of the recent advancements and elucidating the state-of-the-art applications in the field. Furthermore, this review paper serves to highlight potential areas of future investigation by critically evaluating studies from 2019 to 2023, thereby acting as a stepping-stone for further research endeavors.

Identification of small open reading frames in plant lncRNA using class-imbalance learning

Recently, small open reading frames (sORFs) in long noncoding RNA (lncRNA) have been demonstrated to encode small peptides that can help study the mechanisms of growth and development in organisms. Since machine learning-based computational methods are less costly compared with biological experiments, they can be used to identify sORFs and provide a basis for biological experiments. However, few computational methods and data resources have been exploited for identifying sORFs in plant lncRNA. Besides, machine learning models produce underperforming classifiers when faced with a class-imbalance problem. In this study, an alternative method called SMOTE based on weighted cosine distance (WCDSMOTE) which enables interaction with feature selection is put forward to synthesize minority class samples and weighted edited nearest neighbor (WENN) is applied to clean up majority class samples, thus, hybrid sampling WCDSMOTE-ENN is proposed to deal with imbalanced datasets with the multi-angle feature. A heterogeneous classifier ensemble is introduced to complete the classification task. Therefore, a novel computational method that is based on class-imbalance learning to identify the sORFs with coding potential in plant lncRNA (sORFplnc) is presented. Experimental results manifest that sORFplnc outperforms existing computational methods in identifying sORFs with coding potential. We anticipate that the proposed work can be a reference for relevant research and contribute to agriculture and biomedicine.

DMIL-IsoFun: predicting isoform function using deep multi-instance learning

MOTIVATION

Alternative splicing creates the considerable proteomic diversity and complexity on relatively limited genome. Proteoforms translated from alternatively spliced isoforms of a gene actually execute the biological functions of this gene, which reflect the functional knowledge of genes at a finer granular level. Recently, some computational approaches have been proposed to differentiate isoform functions using sequence and expression data. However, their performance is far from being desirable, mainly due to the imbalance and lack of annotations at isoform-level, and the difficulty of modeling gene-isoform relations.

RESULT

We propose a deep multi-instance learning based framework (DMIL-IsoFun) to differentiate the functions of isoforms. DMIL-IsoFun firstly introduces a multi-instance learning convolution neural network trained with isoform sequences and gene-level annotations to extract the feature vectors and initialize the annotations of isoforms, and then uses a class-imbalance Graph Convolution Network to refine the annotations of individual isoforms based on the isoform co-expression network and extracted features. Extensive experimental results show that DMIL-IsoFun improves the Smin and Fmax of state-of-the-art solutions by at least 29.6% and 40.8%. The effectiveness of DMIL-IsoFun is further confirmed on a testbed of human multiple-isoform genes, and Maize isoforms related with photosynthesis.

AVAILABILITY

The code and data are available at http://www.sdu-idea.cn/codes.php?name=DMIL-Isofun.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.