PL-search

PL-search detects remote homologous proteins by constructing profile-links to generate improved position-specific scoring matrices (PSSMs) and hidden Markov models (HMMs) and by scoring sequence pairs with a two-level Jaccard distance to enhance remote homology detection.

Key Features:

• Robust Profile-Link Construction: Builds profile-links using a double-link and iterative extending strategy to produce comprehensive inputs for PSSM or HMM profile construction.
• Two-Level Jaccard Distance Calculation: Calculates similarity scores between sequence pairs using a two-level Jaccard distance metric to provide nuanced measures of sequence similarity.
• Improved Search Performance: Demonstrates improved ranking quality and increases the number of detected remote homologues on widely used benchmark datasets compared with HHblits, JackHMMER, and position-specific iterated-BLAST.

Scientific Applications:

• Protein Structure Prediction: Facilitates detection of remote homologues that inform comparative modeling and structure prediction.
• Functional Annotation: Enhances functional annotation by identifying distant homologues that provide evidence for protein function.
• Evolutionary Studies: Enables evolutionary analyses by uncovering distant evolutionary relationships between proteins.
• Therapeutic Discovery: Aids identification of distant homologues that can inform the development of novel therapeutic strategies.

Methodology:

Constructs profile-links via a double-link and iterative extending strategy, derives PSSMs or HMMs from profile-link information, computes two-level Jaccard distance scores between sequence pairs, and evaluates performance on widely used benchmark datasets.