2lpiRNApred

2lpiRNApred predicts piwi-interacting RNAs (piRNAs) and evaluates their potential to induce target mRNA deadenylation using a two-layer integrated machine learning classifier.

Key Features:

• Two-Layer Classification Framework: Implements a two-stage model that first identifies piRNAs and then predicts their ability to induce target mRNA deadenylation.
• Multi-Strategy Feature Extraction: Extracts sequence features using Kmer, General parallel correlation pseudo-dinucleotide composition, General series correlation pseudo-dinucleotide composition, Normalized Moreau–Broto autocorrelation, and Geary autocorrelation.
• Entropy-Based Feature Selection: Applies Luca fuzzy entropy with Gaussian membership function (LFE-GM) to reduce feature dimensionality and retain informative attributes.
• Dual Machine Learning Classifiers: Uses Sparse Representation Classifier (SRC) and Support Vector Machine with Mahalanobis distance-based radial basis function (SVMMDRBF) to improve prediction accuracy.

Scientific Applications:

• piRNA Identification: Enables computational detection of piwi-interacting RNAs involved in genome defense mechanisms.
• Transposon Silencing Research: Supports investigation of piRNA-mediated silencing of transposable elements.
• Germline Biology Studies: Facilitates analysis of piRNA roles in germ cell formation, germline stem cell maintenance, spermatogenesis, and oogenesis.
• Cancer Genomics: Assists in studying the association between piRNA dysregulation and tumorigenesis.

Methodology:

The method extracts sequence-derived features using multiple composition and autocorrelation descriptors, reduces feature dimensionality using Luca fuzzy entropy with Gaussian membership function (LFE-GM), and performs two-layer classification with Sparse Representation Classifier (SRC) and Support Vector Machine with Mahalanobis distance-based radial basis function (SVMMDRBF).