Recent advances in machine learning methods for predicting LncRNA and disease associations

Abstract

Long non-coding RNAs (lncRNAs) are involved in almost the entire cell life cycle through different mechanisms and play an important role in many key biological processes. Mutations and dysregulation of lncRNAs have been implicated in many complex human diseases. Therefore, identifying the relationship between lncRNAs and diseases not only contributes to biologists’ understanding of disease mechanisms, but also provides new ideas and solutions for disease diagnosis, treatment, prognosis and prevention. Since the existing experimental methods for predicting lncRNA-disease associations (LDAs) are expensive and time consuming, machine learning methods for predicting lncRNA-disease associations have become increasingly popular among researchers. In this review, we summarize some of the human diseases studied by LDAs prediction models, association and similarity features of LDAs prediction, performance evaluation methods of models and some advanced machine learning prediction models of LDAs. Finally, we discuss the potential limitations of machine learning-based methods for LDAs prediction and provide some ideas for designing new prediction models.

FLYNC: A Machine Learning-Driven Framework for Discovering Long Non-Coding RNAs inDrosophila melanogaster

“…The RF model performance metrics are comparable with those of widely used ML-based lncRNA identification models (29,40,41), and other lncRNA-related models (42,43). Even though, such metrics are often not easily comparable since some are calculated with a random balanced subsample of the training data or were tested on different species.…”

Identification of potential LncRNAs as papillary thyroid carcinoma biomarkers based on integrated bioinformatics analysis using TCGA and RNA sequencing data

“…In recent years, machine learning methods for predicting the association between lncRNAs and complex human diseases have become increasingly popular among researchers. Although biological experiments and clinical methods are efficient and reliable, they are time consuming and expensive 31,32 . Computational models can provide the most promising lncRNA-disease associations for further experimental validation, reducing the time and cost of biological experiments 33 .…”

“…This has laid a solid theoretical foundation for lncRNA-disease association prediction research. With further research, related lncRNAs based on machine learning can be used as predictive biomarkers for the treatment and prognosis of glioblastoma, colorectal cancer, lung cancer, bladder cancer, and other tumors 31,[34][35][36] . This also provides us with follow-up research ideas.…”

Identification and coregulation pattern analysis of long noncoding RNAs in the mouse brain after Angiostrongylus cantonensis infection

“…With the flourishing development of high-throughput technologies, the role of lncRNAs in the growth and development of living organisms as well as in disease processes has gradually been revealed. LncRNAs are key genetic regulators of different biological processes and are involved in regulating epigenetic regulation, cell differentiation, the cell cycle, and immune response [ 24 ]. An increasing number of studies have shown that lncRNAs can act as competitive endogenous RNAs that bind to miRNAs and participate in various biological processes [ 25 , 26 ].…”