Adding virtual site support

[IAlibay]

TODO:

• Add end state energy tests!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.12%. Comparing base (5843c69) to head (6603837).

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #59      +/-   ##
==========================================
+ Coverage   95.69%   97.12%   +1.43%     
==========================================
  Files          17       18       +1     
  Lines        1369     1461      +92     
==========================================
+ Hits         1310     1419     +109     
+ Misses         59       42      -17     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[IAlibay]

Benchmarks

Total cost: 11,674 GPU SUs @ UCI HPC3

Results

Here are the results for the Cl and Br benchmarks. Note that these are also uploaded as CSV files in this PR.

Note: all results are in kcal/mol

Cl benchmark results

smiles	pontibus_dG	pontibus_err	gromacs_dG	experiment
c1cc(cc(c1)Cl)N	-4.15	0.05	-4.15	-5.82
c1ccc(cc1)n2c(=O)c(c(cn2)N)Cl	-13.43	0.03	-13.53	-16.43
c1c(c(=O)[nH]c(=O)[nH]1)Cl	-16.11	0.08	-16.22	-17.74
C1[C@@h]2[C@H]3C@@H[C@@]5(C(=C([C@]3(C5(Cl)Cl)Cl)Cl)Cl)Cl	-4.98	0.05	-5.04	-4.82
CCCCCCl	1.29	0.05	1.34	-0.10
CC(C)(C)Cl	0.74	0.04	0.68	1.09
C=CCl	0.70	0.02	0.73	-0.59
C(CCl)OCCCl	-2.94	0.02	-3.11	-4.23
CCOP(=S)(OCC)Oc1c(cc(c(n1)Cl)Cl)Cl	-9.88	0.09	-9.64	-5.04
C(N+[O-])(Cl)(Cl)Cl	-1.49	0.01	-1.48	-1.45
COc1c(c(c(c(c1Cl)C=O)Cl)OC)O	-9.34	0.06	-9.43	-8.68
COP(=O)(C@HO)OC	-12.62	0.63	-12.19	-12.74

Br benchmark results

smiles	pontibus_dG	pontibus_err	gromacs_dG	experiment
c1cc(ccc1Br)Br	-0.75	0.01	-0.74	-2.3
c1cc(ccc1O)Br	-5.65	0.04	-5.70	-5.85
c1c(c(=O)[nH]c(=O)[nH]1)Br	-16.61	0.11	-16.69	-18.17
CBr	0.04	0.04	0.05	-0.82
C(CBr)Cl	-1.47	0.08	-1.69	-1.95
CC(C)(C)Br	0.26	0.01	0.38	0.84
CCCCCBr	0.99	0.03	1.05	-0.1
CCC@Hn1c(=O)c(c([nH]c1=O)C)Br	-12.14	0.01	-12.28	-9.73
C(F)(F)(F)Br	2.24	0.03	2.26	1.79

Analysis

Overall the results agree very well with the Gromacs results Meghan Osato.

The only cases which seem to diverge slightly (i.e. greater than 0.5 kT) is COP(=O)(C@HO)OC with an absolute error of 0.43 kcal/mol from Gromacs. However the standard deviation for three repeats on this solute is 0.63 kcal/mol, so it's likely just a sampling issue.