AGILE

AGILE maps longer sequencing reads (notably 454 reads of approximately 200–1000 bp) to reference genomes using hash table indexing, diagonal multiple seed-match criteria, q-gram filtering, and dynamic incremental search to provide high-throughput, accurate alignments for long-read data.

Key Features:

• Hash Table-Based Algorithm: Employs a hash table-based index for rapid indexing and retrieval of reference positions during mapping.
• Diagonal Multiple Seed-Match Criteria: Identifies potential alignment sites by evaluating multiple seed matches along diagonal paths in the alignment matrix.
• Customized Q-Gram Filtering: Uses tailored q-gram filtering to reduce the search space while maintaining sensitivity and specificity.
• Dynamic Incremental Search Approach: Dynamically adjusts search parameters based on intermediate results to optimize speed and precision.
• Heuristic Optimization: Integrates heuristics across seed identification and alignment scoring to streamline mapping of long reads.
• Performance Comparisons: In comparative studies AGILE was reported to be more accurate than BLAT, comparable to BWA-SW and SSAHA2 in alignment precision, and faster than BLAT and SSAHA2 for reads of 200–1000 bp at practical error rates below 5%.

Scientific Applications:

• Resequencing: Mapping long reads to a reference genome in resequencing studies.
• Variant Detection (SNP identification): Identification of single nucleotide polymorphisms with high precision from long-read alignments.
• Transcript Discovery: Detection of rare transcripts through accurate long-read mapping.
• Long-Read Mapping from 454 Sequencing: Optimized for longer reads generated by 454 sequencing platforms in the ~200–1000 bp range.

Methodology:

The reference is indexed with a hash table; potential alignment sites are identified via diagonal multiple seed-match criteria; customized q-gram filtering and a dynamic incremental search reduce the search space and adjust parameters; final alignments are scored by comprehensive evaluation of candidate matches.