Spectra

makeshift.spectra reads 2D Sparky .ucsf spectra, picks peaks, aligns peak lists, and provides plotting helpers for spectra, peak lists, and chemical-shift perturbations.

nmrglue and NumPy 2.x

This module imports nmrglue. Its current release is incompatible with NumPy 2.x — pin numpy<2 and scipy<1.14 if you hit a data type 'a8' error. See Installation.

Loading a spectrum

Spectrum wraps the intensity array plus one nmrglue unit-conversion object per axis (for ppm ↔ point conversions).

from makeshift.spectra import Spectrum

spec = Spectrum.from_ucsf("hsqc.ucsf")
spec.data           # intensity array (axis 0 = indirect/N, axis 1 = direct/H)
spec.uc             # per-axis unit-conversion objects

Picking peaks

# Noise-floor estimate for setting a sensible baseline
noise = spec.estimate_background()

# Pick peaks in the amide region of a 1H-15N spectrum
peaks = spec.pick_peaks(baseline=noise)

# Convert a point index to ppm on a given axis
ppm = spec.ppm(axis=1, point=512)

pick_peaks accepts an explicit baseline, a peak-picking algorithm, and h_ppm_min / h_ppm_max bounds to restrict the amide window.

Aligning peak lists

map_peaklists aligns two peak lists in (H_ppm, N_ppm) and matches them one-to-one. The right list (e.g. a reference assignment) is shifted by a translation offset — grid-searched if you don't supply one — then Hungarian-matched to the fixed left list within tolerance:

from makeshift.spectra import map_peaklists

left_out, right_mapped = map_peaklists(
    left_peaks.data,          # fixed (e.g. picked from a spectrum)
    right_peaks.data,         # shifted onto left (e.g. reference assignment)
    offset=None,              # (ΔH, ΔN); grid-searched if None
    tol=1.0,                  # multiplier on default (0.03, 0.3) ppm tolerances
    how="inner",
)

left_out carries the transferred labels plus a conflict column flagging right-side peaks that lost a close match to a competitor; right_mapped is the full right list after the offset, ready for plotting.

Plotting

Three helpers, all of which accept an existing Matplotlib ax so you can layer them:

from makeshift.spectra.plotting import plot_spectrum, plot_peaklist, plot_csp

ax = plot_spectrum(spec.data)                 # contour plot
plot_peaklist(peaks.data, ax=ax)              # markers + optional labels
plot_csp(peaks1, peaks2, on="label", ax=ax)   # matched pairs joined by lines

plot_csp draws connecting lines between paired peaks in two matched lists — the standard way to visualise chemical-shift perturbations across a titration or mutation.

Full API

See the Spectra reference.