{
  "slug": "tabix-index",
  "term": "Tabix index",
  "aliases": [
    "TBI index",
    "tabix CSI index"
  ],
  "definition": "A tabix index maps sequence names and genomic intervals to locations in a coordinate-sorted BGZF-compressed tabular file.",
  "note": "Li's 2011 design combines hierarchical bins with a linear index to reduce unnecessary seeks. TBI records include parsing conventions, sequence names and virtual-offset chunks; its index file is itself BGZF compressed. [tabix-paper] [tabix-format]\n\nThe tabix command can create TBI or CSI. TBI has a 2^29-base coordinate limit; CSI generalizes the binning geometry. CSI also serves formats other than tabix text, so not every CSI file is a tabix index. [tabix-manual] [csi-format]\n\nA command-line region such as chr1:101-110 is 1-based inclusive, while a BED interval chr1 100 110 is 0-based half-open. Indexing settings define how input columns are interpreted; the interval query syntax is a separate convention. [tabix-manual]\n\nThe specified TBI fields do not include a mandatory data-file digest. Engineering implication: validate the index against its intended data generation before preserving them together; integrity checks of each file alone do not establish that relationship. [tabix-format]\n\nIn Hesela's synthetic three-record BED experiment, a stale TBI or CSI index returned zero instead of two matching rows with successful exit on HTSlib 1.22.1 and 1.24. Broad and unchanged-region queries passed; rebuilding the index restored the narrow result. An older mtime warned without failing, while equal/newer stale mtimes did not warn. The successful sequential -T control still loaded index metadata. These are designed cases on one host, not production incidence or general index-free recovery. [hesela-pair-evidence]",
  "status": "shipping",
  "advantages": [],
  "limitations": [],
  "supersedes": [],
  "supersededBy": [],
  "firstSeen": null,
  "lastSeen": null,
  "updated": "2026-10-25",
  "related": [
    "bgzf",
    "gzip",
    "backup-manifest"
  ],
  "citations": [
    {
      "id": "tabix-paper",
      "title": "Heng Li. Tabix: fast retrieval of sequence features from generic TAB-delimited files. Bioinformatics 27(5), 718-719 (2011).",
      "url": "https://doi.org/10.1093/bioinformatics/btq671",
      "accessed": "2026-10-25"
    },
    {
      "id": "tabix-manual",
      "title": "HTSlib 1.24 tabix manual: coordinate conventions, TBI limit and CSI selection",
      "url": "https://www.htslib.org/doc/tabix.html",
      "accessed": "2026-10-25"
    },
    {
      "id": "tabix-format",
      "title": "The Tabix index file format: fields, coordinate flag and chunk offsets",
      "url": "https://github.com/samtools/hts-specs/blob/510c107ba89aa13113818c9e4fa3071283cb7f7b/tabix.tex",
      "accessed": "2026-10-25"
    },
    {
      "id": "csi-format",
      "title": "CSI version 1: configurable binning geometry and auxiliary metadata",
      "url": "https://github.com/samtools/hts-specs/blob/510c107ba89aa13113818c9e4fa3071283cb7f7b/CSIv1.tex",
      "accessed": "2026-10-25"
    },
    {
      "id": "hesela-pair-evidence",
      "title": "Hesela: BGZF/tabix generation mismatch, native captures and reproduction method",
      "url": "https://hesela.dev/research/bgzf-tabix-identity/PUBLICATION.md",
      "accessed": "2026-10-25"
    }
  ]
}