History of revisions

A history of revisions of the BGEN format specification is as follows:

BGEN v1.3 (January 2017): link to spec
Support for the zstandard compression library. Tests indicate this has better performance both in terms of file size and speed of reading and writing.
BGEN v1.2 (March 2016): link to spec
Major update extending the BGEN format to add:
• Support for variable ploidy and explicit missing data.
• Support for multi-allelic variants (e.g. complex structural variants).
• Allow for control over file size by supporting genotype probabilities stored at configurable precision.
• Support for storing sample identifiers.
A draft version of this spec was published beginning May 2015. The following changes have been made since the earlier draft:
• 2015-11-05 (v1.2 beta1): modified the treatment of missing data in Layout 2 (v1.2-style) variant data blocks.
• 2016-03-21 (v1.2 beta2): modified the order of stored probabilities for samples with ploidy greater than 2; clarified specification of the phased flag for samples with ploidy less than 2.
BGEN v1.1 (March 2012): link to spec
The first widely used version of the BGEN format. The UK Biobank interim imputed data was released in this format. Relative to v1.0, this version is designed to cope with the long alleles present at indels and structural variants in recent releases of the 1000 genomes project. Features of this version are:
• Support for biallelic SNPs and indels with alleles of arbitrary length (up to 2³²-1).
• Store probabilities to at least 4 decimal places worth' of accuracy
BGEN v1.0 (2009):
The original BGEN format. This version is now unsupported.