Identification of miRNA-Small Molecule Associations by Continuous Feature Representation Using Auto-Encoders

Abstract

MicroRNAs (miRNAs) are short non-coding RNAs that play important roles in the body and affect various diseases, including cancers. Controlling miRNAs with small molecules is studied herein to provide new drug repurposing perspectives for miRNA-related diseases. Experimental methods are time- and effort-consuming, so computational techniques have been applied, relying mostly on biological feature similarities and a network-based scheme to infer new miRNA–small molecule associations. Collecting such features is time-consuming and may be impractical. Here we suggest an alternative method of similarity calculation, representing miRNAs and small molecules through continuous feature representation. This representation is learned by the proposed deep learning auto-encoder architecture. Our suggested representation was compared to previous works and achieved comparable results using 5-fold cross validation (92% identified within top 25% predictions), and better predictions for most of the case studies (avg. of 31% vs. 25% identified within the top 25% of predictions). The results proved the effectiveness of our proposed method to replace previous time- and effort-consuming methods.

Hybrid transformer-CNN model for accurate prediction of peptide hemolytic potential

“…The way we represent peptide sequences profoundly influences the ability of deep-learning models to unlock their hemolytic potential. Automated representation based on deep learning of biological sequences is effective while saving time and effort in traditional methods of gathering information 39 . A thoughtfully designed numerical representation not only captures the essence of each amino acid but also cultivates a structured landscape where patterns of hemolytic activity can emerge.…”

Machine learning in the development of targeting microRNAs in human disease

“…Then, they trained a convolutional neural network to obtain deep retrieved features and adopted the support vector machine classifier to predict latent association. Meanwhile, based on Long Short-Term Memory (LSTM) ( Abdelbaky et al, 2021 ), proposed an encoder-decoder model that could perform on the character level of a sequence. They utilized the LSTM Sequence Auto-Encoders to obtain feature embeddings of miRNAs and small molecules, and sequence-to-sequence learning with an RNN to encode sequences.…”