int32 for indices/indptr

[jianshu93]

Dear biom-format team & Daniel,

In a recent effort to simulate 1 million samples with average sparsity 0.1% per sample (~1 million features) here: https://github.com/jianshu93/sparse_features, I ran into a situation where I can have > 2^32 entries, which will not be supported by biom v2.1 because indices/indptr is by default int32. Hdf5 itself supports int64 so that I do not have problems, but just the biom summary-table and validate commands will fail. Let me know if I misunderstood something. Or any plans for add support for int64 if that's the case.

Best,
Jianshu

[wasade]

The 2.1 format spec is getting dated. Perhaps to time to revise. On Nov 26, 2025, at 14:48, Jianshu_Zhao ***@***.***> wrote:jianshu93 created an issue (biocore/biom-format#999) Dear biom-format team & Daniel, In a recent effort to simulate 1 million samples with average sparsity 0.1% per sample (~1 million features) here: https://github.com/jianshu93/sparse_features, I ran into a situation where I can have > 2^32 entries, which will not be supported by biom v2.1 because indices/indptr is by default int32. Hdf5 itself supports int64 so that I do not have problems, but just the biom summary-table and validate commands will fail. Let me know if I misunderstood something. Or any plans for add support for int64 if that's the case. Best, Jianshu —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>