# Code and data: isotope substitution and the RF cutoff of the FAD–Trp radical pair model

Reproduction archive for the manuscript "Isotope substitution moves the radiofrequency cutoff of the FAD–Trp radical pair model, but not out of the 75–85 MHz band".

Environment used: Python 3.13.5, numpy 2.2.6, scipy 1.17.0 (matplotlib only for the figure). No GPU. Code comments are in Japanese.

## Inputs
- `rf-cutoff-isotope/leb23_fad.json`, `leb23_trp.json`: hyperfine tensors (mT) transcribed from Leberecht et al. 2023 PNAS 120, e2301153120, SI Tables S1 (FAD•−, 15 nuclei) and S2 (TrpH•+, 12 nuclei), flavin principal axis system (x along N5–N10, z normal to the flavin plane). `leb23_tables.py` writes these files from the transcribed MHz values (divided by 28.025 MHz/mT).
- Dipolar tensor: SI Table S3 (MHz), hard-coded as `DT` in `pair_fast.py`.
- `rf-cutoff-isotope/radicalpy_*.json`: alternative DFT tensors distributed with RadicalPy (MIT licence), used only as a cross-check.
- `rf-isotope/c13_tensors_mT.json`: 13C hyperfine tensors of FAD•− from Pažėra et al. 2023 J. Phys. Chem. B 127, 838, SI Table S2. `rf-isotope/numax.py` contains the Leberecht tensors inline.
- Substitution: 1H→2H multiplies tensors by 0.15351 and sets I = 1; 14N→15N multiplies by −1.40275 and sets I = 1/2.

## Commands for each result (run inside the folder named)
| result | folder | command | expected |
|---|---|---|---|
| Table 1, natural Trp span | rf-cutoff-isotope | `python span.py leb23_trp.json 0.05` | 105.53 MHz |
| Table 1, natural FAD span | rf-cutoff-isotope | `python span.py leb23_fad.json 0.05` | 114.92 MHz (about 20 min) |
| Table 1, 2H+15N FAD, 13 nuclei | rf-cutoff-isotope | `python span.py leb23_fad.json 0.05 --sub 1H:D,14N:15N --keep 13` | 62.53 MHz |
| Table 1, zero field | rf-cutoff-isotope | `python span.py leb23_fad.json 0 --sub 1H:D,14N:15N --keep 13` | 61.14 MHz |
| Table 2 | rf-cutoff-isotope | `SUBS=none python pair_fast.py 4 3 1.0` and `SUBS=1H,14N python pair_fast.py 4 3 1.0` | 3.6079e-2 and 1.9376e-3 per state |
| Table 2 | rf-cutoff-isotope | same with `5 3` | 3.522e-2 and 1.947e-3 |
| Sec. 3.3 reference | rf-cutoff-isotope | `SUBS=none python pair_fast.py 5 4 1.0` | 3.363e-2 (6912 states, about 1 GB) |
| Sec. 3.3 single radicals | rf-cutoff-isotope | `SUBS=none python single_spec.py leb23_fad.json 11` / `leb23_trp.json 10` | 1.611e-2 / 2.033e-2 |
| Sec. 3.2 dipolar scaling | rf-cutoff-isotope | `SUBS=1H,14N python pair_fast.py 3 3 0.5 --thr 75` (vary 0.25–1.5) | strength above 75 MHz |
| Figure 1 | rf-cutoff-isotope | `SUBS=none HISTOUT=nat.npz python pair_fast.py 4 3 1.0`, `SUBS=1H,14N HISTOUT=dn.npz python pair_fast.py 4 3 1.0`, `python fig_spectrum.py nat.npz dn.npz` | fig1_spectrum.svg, fig_spectrum_bins.csv |
| Table 2, lower rows (stochastic) | rf-cutoff-isotope | `SUBS=1H,14N python band_trace.py 5 4 --nvec 8`, `SUBS=none python band_trace.py 5 4 --nvec 4`; same with `6 4` (`run_bt_chain.sh`) | 1.908e-3 ± 1.0e-4 and 3.330e-2 ± 9.3e-4 (FAD5·Trp4); 1.883e-3 ± 4.2e-5 and 3.703e-2 ± 4.6e-4 (FAD6·Trp4). FAD6·Trp4 2H+15N takes about 1 h with 7 cores |
| Sec. 2.3 stochastic check | rf-cutoff-isotope | `SUBS=1H,14N python band_trace.py 4 3 --nvec 4` | 1.9381e-3 ± 2.4e-4 (exact 1.9376e-3) |
| Sec. 2.3 error of one vector | rf-cutoff-isotope | `SUBS=1H,14N python band_trace_variance.py 4 3` (also 2 2, 3 2, 3 3) | 0.272 (0.825, 0.472, 1.445) |
| Sec. 2.3 σ and time step | rf-cutoff-isotope | `run_bt_conv.sh` | mean of the same four vectors changes by −0.56% (σ = 0.15 MHz) and −0.04% (half time step) |
| Figure 2 | rf-cutoff-isotope | `SUBS=1H,14N FIG_LANG=en python fig_levels.py` | fig2_levels.svg/png; prints band strength 1.8673e-3 per state and 0.967 for the share of the eight edge-to-edge transitions |
| Sec. 3.3 Eq. 1 weight | rf-isotope | `python same_model_ratio.py 3 2` / `4 3` / `5 2` | 1/13.7, 1/34.0, 1/15.3 |
| Sec. 3.1 13C | rf-isotope | `python scenarios.py FAD7 FAD7_13C_C8 FAD7_13C_C8_C6 FAD7_13C_big3 FAD7_13Call` | +23 MHz for C8, etc. |
| Sec. 3.4 13C band | rf-isotope | `python hist_pair.py "N5,N10,H8,H1'" "HB1,HE1" none` / `"13C:C8"` / `"13C:C8+C6"` | ×6.1, ×20.5 in 116–140 MHz (C8+C6 about 10 min, 2.8 GB) |

The `*.txt` files are the outputs we obtained (dated 2026-11-18/19). On 2026-11-19 the Table 2 values, the Trp span, the FAD3·Trp2 Eq. 1 ratio and the C8 13C shift were re-run from a fresh copy of this archive and agreed with the stored values.

## Licence
Code: MIT. Tensor values are transcriptions of published supplementary tables; cite the original papers when using them.
