PCS

PCS identifies conserved 7-mer k-mers and miRNA seed binding sites in 3′-UTRs by leveraging pairwise evolutionary conservation to detect and prioritize miRNA candidates.

Key Features:

• Conservation-Based Identification: Implements a two-stage strategy using complete base-pairing between the 5'-end of miRNA candidates and potential seed matches within 3′-UTRs to exploit conserved seed matches across species.
• Pairwise Conservation Score (PCS): Computes a PCS for all 7-mers in 3′-UTRs between pairs of Drosophila species, analyzing six pairs of flies to quantify conservation and frequency of 7-mers.
• Machine Learning Integration: Builds a support vector machine (SVM) classifier ensemble named Cons-SVM that uses computed PCSs to identify conserved 7-mers, including miRNA seed matches across known miRNA families.
• High Sensitivity and Performance: Integrates multiple pairwise conservation datasets via machine learning and, in Drosophila melanogaster, identified 689 conserved 7-mers covering 32 of 38 known miRNA families.
• De Novo miRNA Prediction: Searches for 90 nt conserved stem-loop regions containing sequences complementary to conserved 7-mers and analyzes these regions with previously published miRNA prediction software, yielding 47 novel miRNA candidates in a genome-wide screen.

Scientific Applications:

• miRNA Research: Identification of conserved seed binding sites and prediction of novel miRNAs to support studies of miRNA function and posttranscriptional regulation.
• Evolutionary Studies: Use of pairwise conservation scores to examine conservation and functional significance of 7-mers and seed matches across Drosophila species.
• Genomic Screening: Genome-wide screening for conserved 7-mers and stem-loop regions to discover candidate regulatory miRNAs.

Methodology:

Uses a two-stage seed-matching strategy with complete 5'-end base-pairing, computes pairwise conservation scores for all 7-mers across six Drosophila species pairs, trains a Cons-SVM ensemble on PCS features, integrates multiple pairwise conservation datasets, searches for 90 nt conserved stem-loop regions complementary to conserved 7-mers, and evaluates those regions with previously published miRNA prediction software.