Changes in version 0.99.12                       

First Bioconductor release (Bioconductor 3.24). quantMSImageR processes
and quantifies targeted DESI-MRM mass spectrometry imaging data,
extending Cardinal's MSImagingExperiment with a
quant_MSImagingExperiment class that carries calibration and tissue
metadata.

Reading and assembling acquisitions

  - readMRM() reads Waters DESI-MRM acquisitions and names each
    transition from an ion-library CSV, matching precursor and product
    at unit resolution. Isomers the instrument cannot separate are named
    for all of them ("A || B"). Each pixel keeps its stage position.
  - selectTissuePixels() draws the tissue/background mask, and
    labelROIs() draws regions of interest within the tissue (airway_01,
    airway_02, ...). Both are saved beside the acquisition and matched
    by pixel position, so one mask serves every MRM panel of the same
    area.
  - combineMSIs() holds several sections in one object, each its own
    run; bindPanels() merges panels or polarities acquired over the same
    area; and stitchAcquisitions() joins pieces of one tissue by stage
    position. alignFeatures(), trimMSI() and removeBlankMzs() prepare
    them.

Normalisation and filtering

  - int2response() normalises to an internal standard per acquisition
    line, rolling window, sample or pixel, with an ion-library column
    mapping each analyte to its standard when there are several.
  - int2SNR() and applySNR() filter each feature against the
    acquisition's own background pixels (median or mean), with optional
    per-feature thresholds. int2SNR() warns when a mostly-zero
    background leaves the median noise level set by the zero imputation
    rather than by the background. back2NA() and zero2NA() mark
    background and empty values as missing.
  - blockCentre() corrects for a blocking factor -- patient, slide, sex
    -- by dividing each block by its own geometric mean or by its
    baseline arm.

Quantification

  - summariseCalLevels(), createCalCurve() and int2conc() convert
    response to amount (pg per pixel, and pg/mm^2 where the pixel size
    is known) against on-slide ("cal") or on-tissue standard-addition
    ("std_addition") standards, with 1/x, 1/x^2 or no weighting.
  - calibrationDiagnostics(), calibrationR2() and plotCalCoverage()
    report the fit and whether the standards bracket the tissue signal.
    int2conc() stops when more than 10% of a feature's pixels would be
    extrapolated, unless told otherwise.

Visualisation and summaries

  - imageR() draws ion images with capped, histogram-equalised or
    square-root colour scales, in the palettes of quantPalettes().
  - quantileHm() draws per-sample quantile heatmaps. contributionHm()
    colours each cell by the group mean and shades it by the sample's
    contribution to it, so a group effect resting on one replicate is
    visible.
  - createMSIDataMatrix(), tissueData() and tissueMatrix() return tables
    for downstream statistics.

Whole-study workflow

  - runStudy() runs a study from one YAML file: loading, normalisation,
    a sweep of SNR thresholds, optional calibration and ion ratios,
    per-feature .txt images for ImageJ and image registration, and one
    HTML report per threshold. The report holds heatmaps, group
    comparisons, colocalisation, ion images with an animated overview
    (also written as a .gif), regions of interest and an embedded tables
    workbook. On Windows its figures are drawn with the cairo PNG
    device: the default GDI device runs out of graphics handles part-way
    through a large study's ion images.
  - validateConfig() checks a config -- paths, ion-library columns,
    enumerated values, calibration and internal-standard settings -- and
    reports every problem at once. runStudy() will not start on an
    invalid config.
  - generateTxtImages() is the engine behind runStudy(). runExample()
    runs the workflow end to end on the bundled synthetic data: six
    sections in two groups, plus a calibration acquisition.

Changes made during development and review (0.99.0 to 0.99.10) are
listed in full at
https://github.com/MJS-708/quantMSImageR/blob/d1987f2/NEWS.md. Scripts
written against a pre-release version from GitHub need updating for one
rename: function names became camelCase in 0.99.6 (run_study() is now
runStudy(), and so on). Study configs need no changes: the old YAML keys
label: and baseline_label: are still read.