# Complete public whole-exome inventory

Audit date: 2026-08-01

This package is the frozen, machine-readable companion to the LGASieve complete
whole-exome data availability report. Regional slices, gene panels, synthetic
BAMs, spike-ins, chromosome fragments, and controlled-access alignments are not
included in the verified public WES totals.

## Principal results

- Online Dropbox Bioinformatics folder: **0 complete WES files** among 342 BAMs.
- Public complete WES alignment objects with a verified index: **13,948**.
- Indexed alignment volume: **98,884,733,231,276 bytes** (98.885 TB; 89.935 TiB).
- Downloadable complete WES BAMs without a submitted BAI: **895**.
- Public alignment candidates that failed or remained unverified: **35**.
- Separate raw WES cohort: **120 FASTQ files from 60 runs**.
- Unique normalized public alignment objects: **14,878**.
- Conservatively grouped logical alignments: **14,842**.
- Curated or benchmark WES objects: **2,869**.
- Repository-metadata WXS candidates requiring content and coverage QC: **12,009**.

Object counts are not patient counts. Technical replicates, capture-kit
replicates, and reference-build alignments are retained as separate objects.

## Meaning of download verification

Every reported URL was exercised with an anonymous HTTP GET. Indexed objects
were included in the primary total only when both the complete BAM/CRAM and its
BAI/CRAI passed the source audit. The checks used byte-range responses, expected
total sizes, format signatures, and official MD5 values where available.

The entire 98.885 TB corpus was not downloaded. A separate multi-range challenge
tested the start, middle, and tail of seven repository/format representatives:
21/21 requests returned strict HTTP 206 responses with exact ranges, distinct
payload hashes, and valid BAM/CRAM start signatures.

## Main files

- `public_wes_indexed_alignments_verified_2026-08-01.csv`: one row per verified
  alignment-plus-index object.
- `public_wes_unindexed_alignments_verified_2026-08-01.csv`: downloadable BAMs
  that require full download and local indexing.
- `consolidated_public_complete_wes_2026-08-01.csv`: full candidate inventory,
  including failed/unverified rows and alias fields.
- `public_wes_repository_summary_2026-08-01.json`: exact denominators,
  repository summaries, size bins, interpretation, and input/output hashes.
- `dropbox_online_complete_wes_audit_2026-08-01.json`: online Dropbox search
  scope and classification.
- `ena_human_wxs_study_summary_2026-08-01.csv`: 182 ENA study accessions with
  official titles.
- `public_full_wes_brca_specific_alignment_inventory_2026-08-01.csv`: BRCA
  truth annotations and benchmark alignment objects.
- `seqc2_complete_wes_alignment_inventory_verified_2026-08-01.csv`: 24 SEQC2
  tumor/normal WES replicates.
- `public_brca_candidate_raw_wes_inventory_2026-08-01.csv`: 120 raw FASTQ
  objects from PRJNA388048, kept separate from aligned data.
- `public_wes_multirange_spotcheck_2026-08-01.csv/.json`: start/middle/tail
  transfer evidence.
- `SHA256SUMS.txt`: integrity hashes for every packaged file.

## Scientific use

1. Begin with the 2,551 official 1000 Genomes Phase 3 exomes for a consistent
   GRCh37 research cohort.
2. Use NA18949 and HG01528 as independently supported BRCA1 deletion-positive
   challenges. Neither is a duplication positive.
3. Use the 24 SEQC2 files for reproducibility and general CNV stress testing.
4. Inspect BAM/CRAM headers, reference build, capture footprint, callable depth,
   sample type, read groups, duplicates, and contamination before admitting an
   ENA or DepMap object to an LGASieve panel of normals.
5. Use indexed URLs for remote regional access. CRAM requires the exact
   reference. Unindexed BAMs must be downloaded and indexed locally.

## Important limitation

Public availability does not establish clinical truth or clinical performance.
Only two independently supported BRCA deletion-positive people were found, and
no confirmed BRCA duplication-positive complete WES was found. A blinded,
assay-matched, orthogonally confirmed clinical cohort remains required.

## Reproduction

The `scripts` directory contains the enumeration, live-probe, BRCA-source, and
consolidation scripts. Network results are time-dependent; rerunning them later
can change the counts.