DigestedProteinDB: A Compact and Scalable Key-Value Database for In Silico Peptide Digestion and Mass-Based Search

Area of Science:

• Bioinformatics
• Computational Biology
• Mass Spectrometry

Background:

• Efficient comparison of experimental peptide masses with theoretical values is crucial for proteomics, microbial biotyping, and MS imaging.
• These analyses require fast and scalable access to large collections of in silico-digested peptides from evolving protein databases.

Purpose of the Study:

• To present DigestedProteinDB, a compact and high-performance key-value database for peptides generated by enzymatic in silico digestion.
• To optimize disk footprint and accelerate mass-range queries for large-scale peptide data.

Main Methods:

• Implementation using RocksDB, a high-performance key-value store.
• Incorporation of optimization layers: peptide mass discretization and multistage storage compression.
• Benchmarking using 252 million UniProtKB protein sequences, generating 5.9 billion peptides.

Main Results:

• DigestedProteinDB requires ~250 GB disk space and 16 GB system RAM for 5.9 billion peptides.
• Database construction takes ~2 days.
• End-to-end mass-range queries achieve a median latency of ~7 ms per query for 10,000 samples.

Conclusions:

• DigestedProteinDB provides a compact and high-performance resource for mass spectrometry-based applications.
• The database can be used as a standalone resource or integrated into bioinformatics pipelines.
• Modular design allows rapid generation of new databases for various proteases, taxonomic subsets, or digestion parameters.