Mantis: A Fast, Small, and Exact Large-Scale Sequence-Search Index.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Large-scale RNA sequencing data analysis is crucial for understanding gene expression and variation.
• Existing indexing methods like Sequence Bloom Trees (SBTs) have limitations including slow build/query times, inefficient space usage, and false positives.

Purpose of the Study:

• To introduce Mantis, a novel, space-efficient system for indexing and searching raw-read RNA sequencing experiments.
• To overcome the limitations of current sequence search methods.

Main Methods:

• Development of new data structures for efficient indexing of raw sequencing reads.
• Implementation of the Mantis system for large-scale sequence searches.

Main Results:

• Mantis index construction is 6x faster and results in a 20% smaller index compared to the state-of-the-art Split Sequence Bloom Tree (SSBT).
• Mantis query performance is 6-108x faster than SSBT, with zero false positives or negatives.
• Successfully searched 200,400 human transcripts across 2,652 RNA-Seq experiments in 82 minutes, compared to SSBT's ~4 days.

Conclusions:

• Mantis provides a significant advancement in efficiently searching large RNA sequencing datasets.
• The system offers substantial improvements in speed, space efficiency, and accuracy over existing methods.