Changes in version 0.1.0.9001                      

  - Database construction, merging and persistence share typed table and
    value contracts. Conflicting primary identities are refused;
    incomplete links remain explicit. Known legacy empty-table and
    metadata representations are normalized.
  - Actual pathology and XML imports retain patient/sample linkage. VCF
    integration uses explicit nonreference genotypes and preserves
    source fields without inventing annotations. Raw FASTQ/BAM data are
    no longer treated as variant calls.
  - Delimited imports use exact field aliases, support semicolon/tab
    separation, reject malformed outcomes and retain unrecognized source
    columns.
  - VCF ID/INFO/FORMAT/sample values are independent of optional typed
    backends; BAM fields, FASTQ first-n selection and quality values are
    repaired. Malformed input and command failures are visible.
  - Database writes stage and verify serialization, with recovery on
    replacement failure. Loading rejects corrupt objects. Upload
    dispatch preserves the original filename without retaining a deleted
    temporary path.
  - Plot and app filters select the same observations; stale selections
    are cleared and download formats match their contents. Gene survival
    groups describe recorded variants, without inferring a negative
    assay from missing records.
  - Small synthetic cohorts work with stable types; generators restore
    caller RNG and synthetic file dates have an explicit fixed anchor.
  - Development citation metadata no longer advertises the unresolving
    draft DOI; no release or deposit is claimed for this maintenance
    version.

                        Changes in version 0.1.0                        

  - Standardize the named v5 package logo across README, documentation
    pages and available application headers, with matching light and
    dark artwork.

Bug fixes

  - mp_query_variants() now correctly filters by classification and
    variant_type. Previously these arguments were shadowed by the
    columns of the same name during tidy evaluation, so the filters
    silently returned every variant.
  - mp_read_vcf() now returns an empty result for a header-only VCF
    instead of a row of NA values, fixing a reversed index range in the
    text parser.

Features

  - Initial release.
  - Parsers for VCF, FASTQ, BAM, XML reports, PDF reports, Nexus
    pathology/clinical exports, and survival data.
  - In-memory relational database (molpath_db) with build, merge,
    validate, save/load operations.
  - Query engine with tidy evaluation, variant filtering, free-text
    search.
  - Six visualization functions: variant landscape, mutation spectrum,
    survival curves, cohort overview, VAF distribution, patient
    timeline.
  - Interactive Shiny application with 6 tabs and cross-tab
    interactivity.
  - Synthetic data generator (mp_example_db()) for demonstration and
    testing.