Foldcomp

Foldcomp compresses and indexes large collections of protein structures using a lossy structure compression algorithm to reduce storage and enable efficient retrieval for structural biology analyses.

Key Features:

• Lossy structure compression algorithm: Uses a combination of internal and Cartesian coordinates with a bi-directional Neural Radiance Fields (NeRF)-based strategy for compression.
• Improved compression ratio: Achieves approximately a threefold improvement in compression ratio compared to existing methods.
• Reconstruction accuracy: Produces a reconstruction error of 0.08 Å.
• Processing speed: Operates about five times faster than the next fastest compressor while maintaining competitive decompression speeds.
• Indexing system: Provides an index for efficient querying and management of structures by accession numbers.
• Multi-threading: Implements multi-threading to enhance performance on large datasets.

Scientific Applications:

• Structural Biology: Enables storage and analysis of vast structural datasets to support studies of protein structure and function.
• Protein Engineering: Facilitates access to large sets of structural models to aid design and modification of proteins.
• Drug Discovery: Supports exploration of potential drug targets by enabling rapid querying and analysis of protein structures.

Methodology:

Compression combines internal and Cartesian coordinate representations with a bi-directional NeRF-based strategy; an indexing system organizes structures by accession numbers and the implementation uses multi-threading.