Add Charge trying to use local libquadmath

[Elaine Meng]

I can use Chimera addcharge on 1xf6, which has inline MEN (methylasparagine) and LYZ (hydroxylysine), after deleting solvent, ions, and ligands for convenience. However, in ChimeraX, "coulombic" charge calculation dies on the MEN residue, doesn't get to the LYZ:

coulombicUsing Amber 20 recommended default charges and atom types for standard residues
Assigning partial charges to residue MEN (net charge +0) with am1-bcc method
Running ANTECHAMBER command: /Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/antechamber -ek qm_theory='AM1', -i /var/folders/5y/hlccmqph8xj29d001s70mt7r0000gr/T/tmppk6jjdhi/ante.in.mol2 -fi mol2 -o /var/folders/5y/hlccmqph8xj29d001s70mt7r0000gr/T/tmppk6jjdhi/ante.out.mol2 -fo mol2 -c bcc -nc 0 -j 5 -s 2 -dr n
(MEN)
(MEN) Welcome to antechamber 20.0: molecular input file processor.
(MEN)
(MEN) Info: Finished reading file (/var/folders/5y/hlccmqph8xj29d001s70mt7r0000gr/T/tmppk6jjdhi/ante.in.mol2); atoms read (29), bonds read (28).
(MEN) Info: Determining atomic numbers from atomic symbols which are case sensitive.
(MEN) Running: /Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/bondtype -j part -i ANTECHAMBER_BOND_TYPE.AC0 -o ANTECHAMBER_BOND_TYPE.AC -f ac
(MEN)
(MEN)
(MEN) Running: /Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/atomtype -i ANTECHAMBER_AC.AC0 -o ANTECHAMBER_AC.AC -p gaff
(MEN) Info: Total number of electrons: 108; net charge: 0
(MEN)
(MEN) Running: /Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/sqm -O -i sqm.in -o sqm.out
(MEN) dyld: Library not loaded: /usr/local/gfortran/lib/libquadmath.0.dylib
(MEN) Referenced from: /Users/meng/Desktop/ChimeraX_Daily.app/Contents/lib/libgfortran.3.dylib
(MEN) Reason: image not found
(MEN) /Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/antechamber: Fatal Error!
(MEN) Cannot properly run "/Users/meng/Desktop/ChimeraX_Daily.app/Contents/bin/amber20/bin/sqm -O -i sqm.in -o sqm.out".

Tested in UCSF ChimeraX version: 1.2.dev202102121952 (2021-02-12) on mac.

Another structure 6khh containing LYZ but not MEN turned out to be a bad test because in both Chimera and ChimeraX, hydrogens are added incorrectly to that LYZ (treated as C=O carbonyl instead of sp3 CHOH) probably because it doesn't have very good bond lengths.

[Eric Pettersen]

I *believe* this is fixed now -- but it needs testing on a machine without local Fortran libraries installed.

[Elaine Meng]

Confirmed at least for this example: in the daily build, I can successfully use "coulombic" on 1xf6, using Antechamber to calculate charges for the in-polymer nonstandard residues MEN and LYZ. Works with both charge models. Looks almost exactly the same with Gasteiger... perhaps should consider using Gasteiger as default for Coulombic purposes, unless you think it has a higher fail rate.

[Elaine Meng]

On the other hand, Gasteiger wasn't as much faster as I expected, so whichever default you think best.