BGZF

BGZF, short for Block GZip Format, is a gzip-compatible container format designed to provide random access to compressed data. It achieves this by dividing the compressed stream into independently decompressible blocks and by including per-block metadata that supports efficient seeking. Each BGZF block contains a standard gzip wrapper around a DEFLATE-compressed payload, and the uncompressed data in a single block is limited to 64 kilobytes. The blocks are concatenated to form a BGZF file, and the block-level metadata, together with a file-wide index, enables readers to jump directly to a given block without decompressing data preceding it.

To support random access, BGZF files are typically accompanied by an index that maps logical positions to block offsets. A virtual offset combines a block start offset with an offset inside that block, enabling fast random access and partial decompression without processing the entire file. Because the underlying compression within each block uses standard gzip/deflate, BGZF remains compatible with many gzip implementations for full-file decompression, though random access requires BGZF-aware tooling.

Usage and tooling commonly appear in bioinformatics. BGZF is widely used for genomics data formats such as BAM (binary alignment/map) and VCF (variant call format) files that are compressed with BGZF and indexed (for example, by tabix) to support rapid coordinate-based queries. The bgzip program, part of the HTSlib/Samtools ecosystem, creates and decompresses BGZF streams, while libraries like htslib provide APIs for reading BGZF with random access. This design enables efficient storage and querying of large genomic datasets.