CPMG dispersion¶

CPMGExperiment runs a full CPMG relaxation-dispersion pipeline: it reads a set of .ucsf planes, fits peak lineshapes, computes effective R₂ (R₂,eff) per peak across the νCPMG series, classifies which peaks show exchange, and writes a results CSV (with optional plots).

nmrglue and NumPy 2.x

This module reads spectra via nmrglue. Pin numpy<2 and scipy<1.14 if you hit a data type 'a8' error. See Installation.

Running the pipeline¶

The experiment is configured from a YAML file, then run:

from makeshift.relaxation import CPMGExperiment

exp = CPMGExperiment.from_config("SHP2_NSH2_CPMG.yaml")
exp.run(out_dir="out", make_plots=True)

There's also a thin functional wrapper if you prefer a one-liner:

from makeshift.relaxation.cpmg import run_protein

run_protein("SHP2_NSH2_CPMG.yaml", out_dir="out")

The config file¶

The YAML describes the constant-time delay, the reference plane, the sequence, an optional structure for the rigid-body comparison, and one entry per νCPMG plane. An abridged example (see examples/SHP2_NSH2_CPMG.yaml):

time_T2: 0.04            # constant-time CPMG delay, seconds
data_dir: ~/.makeshift/datasets/SHP2_NSH2_CPMG/SHP2_NSH2_CPMG
reference: nuCPMG_0_1_SH2-WB_N15-CPMG.ucsf
peaklist: 52759          # a BMRB id, or a peak-list file
sequence: MTSRRWFHPNITGVEAENLLLTRGVDGSFLARPSKSNPGDF...
pdb: 9EHD
baseline_ref_plane: 50

planes:
  - file: nuCPMG_50_6_SH2-WB_N15-CPMG.ucsf
    vcpmg: 50
  - file: nuCPMG_100_9_SH2-WB_N15-CPMG.ucsf
    vcpmg: 100
  # ... one entry per plane ...

fit_lineshapes_cache: out/caches/SHP2_NSH2_CPMG_fit_lineshapes.json.zip
hydronmr_r2_cache: ../examples/SHP2_NSH2_CPMG_hydronmr_r2.csv

Key	Meaning
`time_T2`	Constant-time CPMG delay (s)
`data_dir`	Directory holding the `.ucsf` planes
`reference`	Reference plane filename
`peaklist`	BMRB id (or file) providing assignments
`sequence`	One-letter sequence
`pdb`	Structure for the rigid-body R₂ comparison
`planes`	List of `{file, vcpmg}` for each dispersion point
`fit_lineshapes_cache`	Cache path for fitted lineshapes (speeds re-runs)
`hydronmr_r2_cache`	Cached HYDRONMR rigid R₂

The example dataset is downloadable — see Datasets & structures:

from makeshift.utils import datasets
datasets.fetch("SHP2_NSH2_CPMG")

Run options¶

run() exposes the pipeline knobs: the lineshape model, an max_r2err_threshold for filtering noisy fits, plotting toggles and axis limits, and a color_map for the classification plots. See the CPMGExperiment.run reference for the full signature.

Outputs¶

run() writes a per-peak results CSV combining assignments, R₂,eff values, and exchange classification to out_dir. With make_plots=True it also produces per-peak dispersion curves, a grid overview, and a sequence waterfall.

Full API¶

See the CPMG reference.