Skip to content

atom Toolbox v3.0.1

Latest

Choose a tag to compare

@mholmboe mholmboe released this 11 Oct 19:17

Official v3.0.1 release of the atom Toolbox.

This release is primarily a correctness pass over the bundled ion force field
parameters. Every ion parameter set in forcefield_functions/ff_params was
audited against its original publication, and the Gromacs ffnonbonded*.itp
files were brought into agreement with them.

Corrected parameters

  • Li & Merz 12-6 IOD, trivalent/tetravalent. The Rmin/2 to sigma column
    had been entered in the paper's row order while the ion names are sorted by
    atomic number, so sigma was attached to the wrong ion for 20 of the 24 ions.
    The set of 24 values was itself correct, so the fix is an exact permutation;
    eps was never affected. Corrected in the four OPC3/OPC/TIP3P-FB/TIP4P-FB IOD
    blocks of every ffnonbonded*.itp and in the 16 affected .mat files.
    Reference: Li, Song & Merz, J. Chem. Theory Comput. 2021, 17, 2342, Table 3.
  • Zhang & Qiu 12-6 ion models. The five anions (F-, F*-, Cl-, Br-, I-) carried
    charge +1 instead of -1, in 110 entries across 22 files.
  • In3+ carried iodine's atomic number and mass (53 / 126.9045) instead of
    indium's (49 / 114.818), in every ffnonbonded*.itp occurrence and 55 .mat files.
  • CLAYFF octahedral Fe. Applied the corrected parameters
    (sigma 0.490620 nm, eps 3.77807E-06 kJ/mol, SI of doi.org/10.1021/acs.jpcc.1c04600)
    to the clayff .mat files and the three lecture clay.ff sets, keeping
    Kerisit's values as comments.
  • ions_Merz_IOD_monovalent_all_ff.mat stored its halide atomtype names with
    U+2212 MINUS SIGN rather than an ASCII hyphen, so they matched no .itp
    atomtype and failed string comparison.

New

  • Zhang/Qiu 12-6 ion sets (_ZQ) added to the ffnonbonded*.itp files for
    11 water models, 594 ion rows each, with atomtype names matching the existing
    Li & Merz (_LM) blocks. Qiu's second fluoride model, written F*- in the paper,
    is named F-_star because * is not safe in a Gromacs atomtype name.
    These parameters assume Lorentz-Berthelot mixing (Gromacs comb-rule 2).
  • GMINFF_k*.mat extended to cover the atomtypes of the MINFF .itp files.
  • MINFF_v1 parameter spreadsheet under ATOM_scripts_lecture/MINFF.

Verified and unchanged

The 18 Li & Merz 12-6-4 sets (328 entries), the four ions_LJ_BL_* sets
(Joung & Cheatham 2008 monovalent plus Babu & Lim 2006 and Aqvist divalent), and
the CM, HFE and divalent IOD sets were all checked against their sources and
needed no change. Unit-variant columns (nm/Angstrom, kJ/kcal/eV/kB, C4, C6/C12)
were checked for mutual consistency across all 130 .mat files.

Note on the very large Zhang/Qiu eps values

The Zhang/Qiu models for highly charged cations deliberately pair a small R
with a very large eps (up to ~1e4 kcal/mol). Those are the published values,
not transcription errors; see the discussion of Figure 5 in
J. Chem. Inf. Model. 2021, 61, 4613.

Fixed file name references

The water model token is lowercase in the actual ff_params filenames, so the
uppercase forms used by the autofit scripts and ff_helpfile2 only resolved on
case-insensitive filesystems. Also corrected a 12-6-4 versus 12_6_4 mismatch
in ff_helpfile, and a commented reference to a CM monovalent set that does not
exist, since Li & Merz published the CM set for divalent ions only.

Earlier v3.x releases

v3, v3.0 and v3.00 are left untouched and still point at their original
2025 commits, so results produced with those parameters remain reproducible.

Versioning

The toolbox version is 3.0.1, following MATLAB's Major.Minor.Patch
identifier requirement. Every function's %% Version header, the published
HTML documentation and CITATION.cff were updated accordingly.