Improved Search of Large Transcriptomic Sequencing Databases Using Split Sequence Bloom Trees.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Vast RNA-sequencing (RNA-seq) datasets, such as those in the NIH Sequencing Read Archive, offer immense potential for biological discovery.
• Current limitations in searching these large-scale datasets hinder the extraction of valuable information on condition-specific expression and population variation.

Purpose of the Study:

• To develop and evaluate a novel indexing scheme for efficient sequence-based querying of terabyte-scale RNA-seq data collections.
• To improve upon existing data structures like Sequence Bloom Trees (SBTs) for handling massive sequencing experiment data.

Main Methods:

• Introduction of split sequence bloom trees (SSBTs), an enhanced data structure for indexing short-read sequencing data.
• Application of SSBTs to query expression patterns of specific transcripts across a large dataset of 2652 RNA-seq experiments (breast, blood, brain tissues).

Main Results:

• SSBTs provide a fivefold improvement in search and storage costs compared to SBTs.
• Querying a 1000-nucleotide sequence within the indexed dataset takes under 4 minutes on a single thread.
• The SSBT index for the entire dataset requires only 39 GB of storage.

Conclusions:

• SSBTs offer a scalable and efficient solution for sequence-based querying of large RNA-seq experiment collections.
• This methodology significantly enhances the utility of public sequencing data archives for biological research.
• The developed indexing scheme facilitates faster identification of gene expression conditions and variations.