Extract Matrix of aligned Signal Intensities

Extracts a peak-area matrix from aligned spectra. Rows are chemical-shift positions, columns are spectra. When maxCombine = 0 the raw aligned integral vectors are returned. When maxCombine > 0, peaks are combined using one of two methods controlled by combineMethod.

Usage

get_si_mat(x, ref = NULL, drop_zero = FALSE, maxCombine = 0, combineMethod = 2)

Arguments

x

An object of type aligns.

ref

A single align or decon2 object whose peaks define the rows of the output matrix. If NULL (default), the reference is auto-detected from x via find_ref(). Used by combineMethod = 2 only.

drop_zero

If TRUE, rows where all values are zero are removed.

maxCombine

Controls peak combining in datapoints. 0 (default): off. Any positive number enables peak combining. For combineMethod = 1 this is the range parameter of combine_peaks(). For combineMethod = 2 this is the maximum distance to the nearest reference peak. See 'Details'.

combineMethod

Peak-combining backend. 1: merge adjacent columns via combine_peaks(). 2 (default): nearest-neighbour snapping to reference peaks.

Value

A numeric matrix with chemical shifts as rownames and spectrum names as colnames.

Details

Method 1 (combineMethod = 1) passes the transposed integral matrix to combine_peaks() with range = maxCombine. Partly-filled neighbouring columns are merged when they share no common non-zero row.

Method 2 (combineMethod = 2) maps every peak to the nearest reference peak and keeps it only if the distance is at most maxCombine datapoints. Areas of peaks that map to the same reference peak are summed.

Example for method 2 with ref peaks at indices 5, 9, 20 and maxCombine = 4:

Step 1 – Build intervals [ref ± maxCombine]:

  ref:     5              9                    20
           |              |                     |
  int:  [1 ····· 9]   [5 ···· 13]        [16 ···· 24]
            overlap!

Step 2 – Shrink overlapping neighbours to midpoint.
         Refs 1 & 2 overlap → mid = floor((5+9)/2) = 7.
         Refs 2 & 3 don't  → keep maxCombine boundary.

  ref:     5         9                         20
           |         |                          |
  int:  [1 ··· 7] [8 ·· 13]   gap        [16 ···· 24]

Step 3 – Assign peaks to nearest ref; keep if ≤ maxCombine:

  | Peak | Nearest | Dist | ≤ 4? | Action |
  |------|---------|------|------|--------|
  |    3 | ref 1   |    2 | yes  | keep   |
  |    6 | ref 1   |    1 | yes  | keep   |
  |    8 | ref 2   |    1 | yes  | keep   |
  |   11 | ref 2   |    2 | yes  | keep   |
  |   15 | ref 3   |    5 | no   | DROP   |
  |   21 | ref 3   |    1 | yes  | keep   |

Step 4 – Sum areas (A × pi) per reference peak:

  ref 1 ← A(3) + A(6)    (peaks at 3, 6)
  ref 2 ← A(8) + A(11)   (peaks at 8, 11)
  ref 3 ← A(21)          (peak 15 dropped)

Author

2024-2025 Tobias Schmidt: initial version.

Examples

if (interactive()) {
    decons <- deconvolute(sim[1:2], sfr = c(3.55, 3.35))
    aligns <- align(decons, maxCombine = 0)
    si_mat_0 <- get_si_mat(aligns)
    si_mat_1 <- get_si_mat(aligns, maxCombine = 20)
    si_mat_2 <- get_si_mat(aligns, maxCombine = 20, combineMethod = 1)
}