Inspection of iodine containing molecules in 2nd gen. training set for valence parameter optimization

• In 2nd gen training set (used for Parsley version>=1.2.0), 40 optgeo, 5 td, 3 vibfreq targets have iodine. For optimized geometry, 10 unique molecules Cc1c(cnn1CC(=O)C)I,  c1cc(cc(c1)I)[C@@H]2CCN2, c1cc(ccc1N/N=C(/C#N)\\c2nnc3n2CCCCC3)I, CC(C)(C#N)C(=O)Nc1cc(ccc1Cl)I, c1cc(nc(c1)I)SC(F)(F)F, c1cc(c(c(c1)I)O)C(=O)O, c1cc(c(cc1Br)I)[N+](=O)[O-],  Cc1c(cn(n1)CC(=O)C)I,  Cc1cc(nc(n1)N/N=C/c2c(nc3n2cc(cc3)I)C)C,  Cc1c(cc(c(c1Cl)I)Cl)[N+](=O)[O-] were selected from eMolecules discrepancy set. All of them are molecules with an iodine attached to an aromatic ring.

• All parameters assigned to substructures including an iodine

  • Bonds

    • ID: b73, smirks: [#6:1]-[#53:2]

  • Angles

    • ID: a10, smirks: [*:1]~[#6X3:2]~[*:3]

    • ID: a14, smirks: [*:1]~;!@[*;X3;r5:2]~;@[*;r5:3]

  • Torsions

    • ID: t43, smirks: [*:1]~[#6X3:2]-[#6X3:3]~[*:4]

    • ID: t44, smirks: [*:1]~[#6X3:2]:[#6X3:3]~[*:4]

    • ID: t45, smirks: [*:1]-,:[#6X3:2]=[#6X3:3]-,:[*:4]

    • ID: t47, smirks: [*:1]~[#6X3:2]-[#6X3$(*=[#8,#16,#7]):3]~[*:4]

    • ID: t75, smirks: [*:1]~[#7X2,#7X3$(*~[#8X1]):2]:[#6X3:3]~[*:4]

  • Angle and torsion terms assigned to the iodine-containing substructures are general terms, not specific for iodine containing substructures(terminal atoms are not specified to iodine). And the torsion terms are described by a 2-fold cosine function only(f(x)=k(1+cos(x-180))), forcing all MM optimized geometries to locate iodine and the connected aromatic ring on the same plane, regardless of the reference QM optimized geometries.

  • For bonds terms, specific rules are defined for bonds connecting iodine to another element. Therefore, the iodine-related bonds, fitted to the wrong QM geometries, need to be re-fitted. (initial guess on the equilibrium length of b73= 2.075 Å, v1.2.0=2.25 Å, bond lengths at QM optimized geometries= 2.19 ~ 2.33 Å)

• For example,

  • targets/optgeo_OpenFF_Gen_2_Opt_Set_4_eMolecules_Discrepancy-01/52_C7H9IN2O

    • reference QM (DZVP) vs MM from v1.2.0 (transparent red)

      Open

      

    • QM w/ DZVP vs QM w/ def2-svp (diffuse blue)

      • Open

        

  • targets/optgeo_OpenFF_Gen_2_Opt_Set_4_eMolecules_Discrepancy-15/778_C9H10IN

    • reference QM w/ DZVP vs MM from v1.2.0

       

TODO

• Add iodine containing molecules to the theory benchmarking set

•