HYDRONMR prediction¶

makeshift.hydronmr is a pure-Python port of Fast-HYDRONMR: it predicts per-residue NMR relaxation (T1, T2, T1/T2, NOE) from a protein PDB structure by computing a rigid-body diffusion tensor. It's the engine behind the rigid-body comparison in a relaxation profile.

Quick start¶

from makeshift.hydronmr import run

result = run("in.pdb")

run() takes two optional arguments:

config_path= — a different config.yml (defaults to the one bundled next to the engine).
csv_path= — also write a CSV (resseq, T1, T2, T1_over_T2, NOE).

Getting results¶

The Result object exposes tidy views — no intermediate file needed:

result = run("in.pdb")

df = result.to_dataframe()
#   chain  seqpos    T1    T2  T1_over_T2   NOE
#   one row per residue

result.to_csv("t1t2.csv")     # resseq, T1, T2, T1_over_T2, NOE

The raw mapping is also available:

result.per_residue          # {(chain, resseq): (T1, T2, T1_over_T2, NOE)}

to_dataframe() is the general, correctly-labelled table; derive whatever a given analysis needs from it.

Where the structure comes from¶

run() takes a local PDB path. To fetch a structure by PDB id or UniProt accession first, use fetch_structure:

from makeshift.utils import fetch_structure

pdb = fetch_structure("1WRP")        # RCSB
pdb = fetch_structure("P0DP23")      # AlphaFold DB
result = run(pdb)

Full API¶

See the HYDRONMR reference.