A BAM file is a compressed binary format used to store sequencing read alignments to a reference genome. A BAI file is the index that allows rapid retrieval of alignments from specific genomic regions within a BAM file.
Together, they form the standard representation of aligned sequencing data used in most genomics pipelines.
Why BAM files exist
Alignment output in SAM format is large and slow to process because it is plain text. Whole-genome sequencing can produce billions of reads, making direct storage and analysis inefficient.
BAM stores the same information in compressed binary form, greatly reducing size and enabling fast region-based access when an index is present.
This makes genome-scale analysis and interactive visualisation practical.
Core structure of a BAM file
A BAM file contains:
- A header describing reference sequences and processing history.
- Alignment records for sequencing reads mapped to those references.
- Block-compressed data that allows indexed random access.
Each alignment record corresponds to a read and includes its mapping location, alignment operations, quality values, and status flags.
BAM files are typically coordinate-sorted before indexing so that reads appear in genomic order.
Example alignment record
Because BAM is binary, alignments are commonly viewed through their SAM representation:
r001 99 chr1 10001 60 50M = 10100 149 ACTG... FFFF...
Key elements:
- Read
r001maps to chromosome 1 at position 10001. - Mapping quality indicates confidence in placement.
- The CIGAR string describes how the read aligns.
- Flags encode pairing, orientation, and mapping status.
BAM stores this information in compressed binary form rather than text.
Details that matter
- BAM must be coordinate-sorted before indexing.
- A
.baiindex enables rapid access to selected regions. - Without an index, tools must scan the entire file.
- BAM internally uses 0-based coordinates even though SAM displays 1-based positions.
- Mapping status is determined by alignment flags, not just coordinates.
Common interpretation mistakes
Frequent operational errors include:
- Attempting region queries without a BAI index.
- Indexing an unsorted BAM file.
- Assuming reads with coordinates are always mapped.
- Confusing internal coordinate systems when comparing formats.
Where BAM fits in pipelines
Typical workflow:
Sequencing → Alignment → BAM → Variant calling → Interpretation
Most downstream tools operate directly on BAM files.
BAM and CRAM
CRAM is a newer alignment format that compresses data more efficiently by storing differences relative to a reference genome.
BAM remains widely used because it is simple, robust, and broadly supported, while CRAM is increasingly adopted for large-scale storage.
Technical specification
The official SAM, BAM, and BAI specifications are maintained in the hts-specs repository:
https://samtools.github.io/hts-specs/
