EmiRPred

EmiRPred predicts exosomal and non-exosomal microRNAs (miRNAs) to distinguish miRNA localization and support biomarker discovery for liquid biopsy.

Key Features:

• Predictive Models: Integration of alignment-based and AI-based predictive models for exosomal miRNA identification.
• Alignment-Based Approaches: Motif-based MERCI and similarity-based BLAST methods are used, with reported coverage of approximately 29%.
• Artificial Intelligence Models: Machine learning (ML), deep learning (DL), and large language models (LLMs) were employed, achieving up to AUC 0.707 and MCC 0.268 on independent datasets.
• Ensemble Method: An ensemble combining alignment-based and AI-based approaches improves performance, achieving up to AUC 0.73 and MCC 0.352.
• Dataset: Models were developed using 956 exosomal and 956 non-exosomal miRNA sequences sourced from RNALocate and miRBase.
• Motif Identification: Motif recognition for nucleotide patterns associated with exosomal miRNAs is incorporated via motif-based analysis.

Scientific Applications:

• Liquid Biopsy Biomarkers: Identification of exosomal miRNAs as potential non-invasive biomarkers for liquid biopsy research and biomarker discovery.
• Motif Identification: Recognition of specific nucleotide motifs associated with exosomal miRNAs to inform design and prediction of these molecules.

Methodology:

Built using 956 exosomal and 956 non-exosomal miRNA sequences from RNALocate and miRBase; applied MERCI motif-based and BLAST similarity-based alignment approaches, machine learning, deep learning, and large language models; combined alignment- and AI-based predictions in an ensemble; reported evaluation metrics AUC and MCC on independent datasets, with alignment approaches showing ~29% coverage.