ENH: support GMX linear angle potential (#5372)

* ENH: support GMX linear angle potential

* Fixes gh-5361.
* Add support for parsing linear angle potentials in GMX TPR files. As described in the GMX docs linked in the matching issue, these are useful for planar molecules like CO2.
* TPR position and angle data-related regression tests have been added. I did check that the angle data test fails before the patch applied here.
* I believe there may be a serious unit/scaling issue in the TPR coordinate parser, but I'll open a separate ticket about that since it is not directly related.
* Update CHANGELOG

Author: tylerjereddy

package/CHANGELOG

@@ -60,6 +60,8 @@ Fixes
    DSSP by porting upstream PyDSSP 0.9.1 fix (Issue #4913)
 
 Enhancements
+ * Read Linear Angle bonded interaction entries from GROMACS TPR files and 
+   add the angles to the topology (Issue #5361, PR #5372)
  * Enables parallelization for analysis.atomicdistances.AtomicDistances
    (Issue #4662, PR #4822)
  * DSSP uses ``capped_distance`` and ``calc_bonds`` for faster distance

package/MDAnalysis/topology/tpr/utils.py

@@ -826,7 +826,7 @@ def do_moltype(data, symtab, fver):
                 bonds += list(ik_obj.process(ias))
             elif ik_obj.name in ['ANGLES', 'G96ANGLES', 'CROSS_BOND_BOND',
                                  'CROSS_BOND_ANGLE', 'UREY_BRADLEY', 'QANGLES',
-                                 'RESTRANGLES', 'TABANGLES']:
+                                 'RESTRANGLES', 'TABANGLES', 'LINEAR_ANGLES']:
                 angles += list(ik_obj.process(ias))
             elif ik_obj.name in ['PDIHS', 'RBDIHS', 'RESTRDIHS', 'CBTDIHS',
                                  'FOURDIHS', 'TABDIHS']:

testsuite/MDAnalysisTests/coordinates/test_tpr.py

@@ -68,6 +68,7 @@
     TPR2020B2,
     INPCRD,
     TPR_gh_5145,
+    TPR_linear_angle,
 )
 import MDAnalysis as mda
 from MDAnalysis.coordinates.TPR import TPRReader
@@ -84,6 +85,16 @@
         # NOTE: expected values are expressed in nm
         # units and converted to Angstrom in the body
         # of the test below, before assertions
+      
+        # see gh-5361 for CO2/linear angle:
+        (
+            TPR_linear_angle,
+            [1.250, 1.250, 1.250],
+            [1.250, 1.260, 1.136],
+            (3, 3),
+            [0, 0, 0],
+            [0, 0, 0],
+        ),      
         # this case is an alanine dipeptide
         # with neural network potential active
         # and nonzero velocities

testsuite/MDAnalysisTests/data/tprs/Linear_Angle.tpr

@@ -137,6 +137,7 @@
     "TPR2024_4",
     "TPR2025_0",
     "TPR2026_0",
+    "TPR_linear_angle",
     "TPR510_bonded",
     "TPR2016_bonded",
     "TPR2018_bonded",
@@ -574,6 +575,7 @@
 TPR2024_4 = (_data_ref / "tprs/2lyz_gmx_2024_4.tpr").as_posix()
 TPR2025_0 = (_data_ref / "tprs/2lyz_gmx_2025_0.tpr").as_posix()
 TPR2026_0 = (_data_ref / "tprs/2lyz_gmx_2026_0.tpr").as_posix()
+TPR_linear_angle = (_data_ref / "tprs/Linear_Angle.tpr").as_posix()
 # double precision
 TPR455Double = (_data_ref / "tprs/drew_gmx_4.5.5.double.tpr").as_posix()
 TPR460 = (_data_ref / "tprs/ab42_gmx_4.6.tpr").as_posix()

testsuite/MDAnalysisTests/datafiles.py

@@ -51,7 +51,8 @@
                                         TPR2023_bonded, TPR2024_4_bonded,
                                         TPR2025_0_bonded, TPR2024_bonded,
                                         TPR2026_0_bonded,
-                                        TPR_NNPOT_2025_0, TPR_NNPOT_2026_0)
+                                        TPR_NNPOT_2025_0, TPR_NNPOT_2026_0,
+                                        TPR_linear_angle)
 from numpy.testing import assert_equal
 
 # fmt: on
@@ -438,3 +439,12 @@ def test_resids(resid_from_one, resid_addition):
         resids,
         err_msg="tpr_resid_from_one kwarg not switching resids",
     )
+
+
+def test_gh_5361():
+    # GROMACS linear angle potential handling with CO2 example
+    u = mda.Universe(TPR_linear_angle)
+    actual = u.angles.to_indices()
+    expected = [[1, 0, 2]]
+    assert_equal(u.atoms.names, ["C", "O1", "O2"])
+    assert_equal(actual, expected)