Merge pull request #64 from Erastova-group/siminp_fixes

Add updated core and UC files

package/ClayCode/core/classes.py

@@ -2007,6 +2007,13 @@ def __init__(self, path: Union[BasicPath, Dir], check=False, order=None):
         #     self._data.append(FileFactory(file, check=check))
         self.data = self._data
 
+    def __repr__(self):
+        return (
+            f"{self.__class__.__name__}({self.path.name}: "
+            + ", ".join([file.name for file in self.data])
+            + ")"
+        )
+
 
 class PathList(FileList):
     pass

package/ClayCode/core/consts.py

@@ -39,3 +39,5 @@
 TABSIZE = 4
 
 ANGSTROM = "\u212B"
+GREATER_EQUAL = "\u2265"
+LESS_EQUAL = "\u2264"

package/ClayCode/core/lib.py

@@ -167,6 +167,7 @@ def run_analysis(instance: analysis_class, start: int, stop: int, step: int):
         if v is not None:
             kwarg_dict[k] = v
     instance.run(**kwarg_dict)
+    return instance
 
 
 @overload
@@ -217,44 +218,77 @@ def get_selections(infiles, sel, clay_type, other=None, in_memory=False):
     u = MDAnalysis.Universe(*infiles, in_memory=in_memory)
     # only resname specified
     if len(sel) == 1:
+        sel_str = sel[0]
         sel = u.select_atoms(f"resname {sel[0]}")
+
     # rename and atom type specified
     elif len(sel) == 2:
         # expand search string for terminal O atom types
         if sel[1] == "OT*":
             sel[1] = "OT* O OXT"
+        sel_str = "{}: {}".format(*sel)
         sel = u.select_atoms(f"resname {sel[0]}* and name {sel[1]}")
+
     else:
         raise ValueError('Expected 1 or 2 arguments for "sel"')
+
     if other is None:
         pass
     elif len(other) == 1:
         logger.debug(f"other: {other}")
+        other_str = other[0]
         other = u.select_atoms(f"resname {other[0]}")
+
     elif len(other) == 2:
         logger.debug(f"other: {other}")
         if other[1] == "OT*":
             other[1] = "OT* O OXT"
+        other_str = "{}: {}".format(*other)
         other = u.select_atoms(f"resname {other[0]}* and name {other[1]}")
     else:
         raise ValueError('Expected 1 or 2 arguments for "other"')
     clay = u.select_atoms(f"resname {clay_type}* and name OB* o*")
-    logger.finfo(
-        f"'clay': Selected {clay.n_atoms} atoms of "
-        f"{clay.n_residues} {clay_type!r} unit cells"
+    # logger.finfo(
+    #     f"'clay': Selected {clay.n_atoms} atoms of "
+    #     f"{clay.n_residues}  unit cells"
+    # )
+    log_atomgroup_info(
+        ag=clay, ag_name=clay_type, kwd_str="Clay unit cell type"
     )
 
     sel = select_outside_clay_stack(sel, clay)
     # Clay + two other atom groups selected
+    log_atomgroup_info(ag=sel, ag_name=sel_str, kwd_str="Atom selection")
+
     if other is not None:
         other = select_outside_clay_stack(other, clay)
+        log_atomgroup_info(
+            ag=other, ag_name=other_str, kwd_str="Second atom selection"
+        )
         return sel, clay, other
 
     # Only clay + one other atom group selected
     else:
         return sel, clay
 
 
+def get_ag_numbers_info(ag: AtomGroup) -> Tuple[int, int]:
+    return "Selected {} atoms in {} residues.".format(
+        ag.n_atoms, ag.n_residues
+    )
+
+
+def log_atomgroup_info(ag: AtomGroup, kwd_str: str, ag_name: str):
+    if ag.n_atoms > 0:
+        logger.finfo(f"{ag_name!r}", kwd_str=f"{kwd_str}: ")
+        logger.finfo(get_ag_numbers_info(ag))
+    else:
+        logger.error(
+            f"{ag_name} contains 0 atoms. Check atom selection parameters."
+        )
+        sys.exit(1)
+
+
 def select_clay(
     universe: MDAnalysis.Universe,
     ff: Optional[ForceField] = None,
@@ -568,7 +602,7 @@ def select_cyzone(
     :rtype: NoReturn
     """
     z_col = distances[:, :, 2]
-    z_col.mask = np.abs(distances[:, :, 2]) > z_dist
+    z_col.mask = np.abs(z_col) > z_dist
     distances.mask = np.broadcast_to(
         z_col.mask[:, :, np.newaxis], distances.shape
     )

package/ClayCode/core/parsing.py

@@ -275,6 +275,127 @@
     "analyse", help="Analyse clay simulations."
 )
 
+analysisparser.add_argument(
+    "-name", type=str, help="System name", dest="sysname", required=True
+)
+analysisparser.add_argument(
+    "-inp",
+    type=str,
+    help="Input file names",
+    nargs=2,
+    metavar=("coordinates", "trajectory"),
+    dest="infiles",
+    required=False,
+)
+analysisparser.add_argument(
+    "-inpname",
+    type=str,
+    help="Input file names",
+    metavar="name_stem",
+    dest="inpname",
+    required=False,
+)
+analysisparser.add_argument(
+    "-uc",
+    type=str,
+    help="Clay unit cell type",
+    dest="clay_type",
+    required=True,
+)
+analysisparser.add_argument(
+    "-sel",
+    type=str,
+    nargs="+",
+    help="Atom type selection",
+    dest="sel",
+    required=True,
+)
+analysisparser.add_argument(
+    "-n_bins",
+    default=None,
+    type=int,
+    help="Number of bins in histogram",
+    dest="n_bins",
+)
+analysisparser.add_argument(
+    "-bin_step",
+    type=float,
+    default=None,
+    help="bin size in histogram",
+    dest="bin_step",
+)
+analysisparser.add_argument(
+    "-xyrad",
+    type=float,
+    default=3,
+    help="xy-radius for calculating z-position clay surface",
+    dest="xyrad",
+)
+analysisparser.add_argument(
+    "-cutoff",
+    type=float,
+    default=20,
+    help="cutoff in z-direction",
+    dest="cutoff",
+)
+
+analysisparser.add_argument(
+    "-start",
+    type=int,
+    default=None,
+    help="First frame for analysis.",
+    dest="start",
+)
+analysisparser.add_argument(
+    "-step",
+    type=int,
+    default=None,
+    help="Frame steps for analysis.",
+    dest="step",
+)
+analysisparser.add_argument(
+    "-stop",
+    type=int,
+    default=None,
+    help="Last frame for analysis.",
+    dest="stop",
+)
+analysisparser.add_argument(
+    "-out",
+    type=str,
+    help="Filename for results pickle.",
+    dest="save",
+    default=True,
+)
+analysisparser.add_argument(
+    "-check_traj",
+    type=int,
+    default=False,
+    help="Expected trajectory length.",
+    dest="check_traj_len",
+)
+analysisparser.add_argument(
+    "--write_z",
+    type=str,
+    default=True,
+    help="Binary array output of selection z-distances.",
+    dest="write",
+)
+analysisparser.add_argument(
+    "--overwrite",
+    action="store_true",
+    default=False,
+    help="Overwrite existing z-distance array data.",
+    dest="overwrite",
+)
+analysisparser.add_argument(
+    "--update",
+    action="store_true",
+    default=False,
+    help="Overwrite existing trajectory and coordinate array data.",
+    dest="new",
+)
+
 # plot analysis results
 plotparser = subparsers.add_parser(
     "plot", help="Plot simulation analysis results"
@@ -309,6 +430,7 @@
     dest="yaml_file",
 )
 
+
 # siminp_subparsers = siminpparser.add_subparsers()
 
 # dspace_arg_group = siminpparser.add_argument_group("il_spacing")
@@ -703,7 +825,9 @@ def check(self) -> None:
                 )
                 if tbc_match:
                     pass
-                self.uc_stem = self._uc_path.itp_filelist[0].stem[:-3]
+                self.uc_stem = self._uc_path.itp_filelist.filter(
+                    "[A-Z][A-Z0-9]\d\d\d"
+                )[0].stem[:-3]
                 logger.debug(f"Setting unit cell type: {self._uc_name!r}")
             else:
                 raise ValueError(f"Unknown unit cell type {uc_type!r}!")

package/ClayCode/data/data/UCS/D21/D2000.gro

@@ -0,0 +1,43 @@
+trans-vacant dioctahedral smectite, Garfield Nontronite, AMCSD: 0007180
+40
+    1D200   HO1    1   0.183   0.000   0.583
+    1D200   HO2    2   0.447   0.457   0.583
+    1D200   HO3    3   0.158   0.000   0.377
+    1D200   HO4    4   0.422   0.457   0.377
+    1D200   OH1    5   0.084   0.000   0.583
+    1D200   OH2    6   0.348   0.457   0.583
+    1D200   OH3    7   0.257   0.000   0.377
+    1D200   OH4    8  -0.007   0.457   0.377
+    1D200   OB1    9   0.049   0.285   0.590
+    1D200   OB2   10   0.312   0.742   0.590
+    1D200   OB3   11   0.049   0.629   0.590
+    1D200   OB4   12   0.312   0.172   0.590
+    1D200   OB5   13   0.292   0.285   0.370
+    1D200   OB6   14   0.029   0.742   0.370
+    1D200   OB7   15   0.292   0.629   0.370
+    1D200   OB8   16   0.029   0.172   0.370
+    1D200   OB9   17   0.009   0.457   0.796
+    1D200  OB10   18   0.272   0.000   0.796
+    1D200  OB11   19   0.333   0.457   0.164
+    1D200  OB12   20   0.069   0.000   0.164
+    1D200  OB13   21   0.401   0.686   0.814
+    1D200  OB14   22   0.137   0.229   0.814
+    1D200  OB15   23   0.401   0.229   0.814
+    1D200  OB16   24   0.137   0.686   0.814
+    1D200  OB17   25   0.468   0.686   0.146
+    1D200  OB18   26   0.204   0.229   0.146
+    1D200  OB19   27   0.468   0.229   0.146
+    1D200  OB20   28   0.204   0.686   0.146
+    1D200  FEO1   29   0.171   0.152   0.480
+    1D200  FEO2   30   0.434   0.609   0.480
+    1D200  FEO3   31   0.171   0.762   0.480
+    1D200  FEO4   32   0.434   0.305   0.480
+    1D200   ST1   33   0.017   0.301   0.753
+    1D200   ST2   34   0.281   0.758   0.753
+    1D200   ST3   35   0.017   0.613   0.753
+    1D200   ST4   36   0.281   0.156   0.753
+    1D200   ST5   37   0.324   0.301   0.207
+    1D200   ST6   38   0.060   0.758   0.207
+    1D200   ST7   39   0.324   0.613   0.207
+    1D200   ST8   40   0.060   0.156   0.207
+   0.52770   0.91400   0.96003   0.00000   0.00000   0.00000   0.00000  -0.18661   0.00000

package/ClayCode/data/data/UCS/D21/D2000.itp

@@ -0,0 +1,57 @@
+;
+;
+[ moleculetype ]
+; name		nrexcl
+  D2000		1
+
+[ atoms ]
+;   nr       type  resnr residue  atom   cgnr     charge       mass  typeB    chargeB      massB
+; residue   1 D2000rtp D2000q  0.0
+     1         ho      1    D2000 HO1      1      0.425      1.008   
+     2         ho      1    D2000 HO2      2      0.425      1.008   
+     3         ho      1    D2000 HO3      3      0.425      1.008   
+     4         ho      1    D2000 HO4      4      0.425      1.008   
+     5         oh      1    D2000 OH1      5      -0.95         16   
+     6         oh      1    D2000 OH2      6      -0.95         16   
+     7         oh      1    D2000 OH3      7      -0.95         16   
+     8         oh      1    D2000 OH4      8      -0.95         16   
+     9         ob      1    D2000 OB1      9      -1.05         16   
+    10         ob      1    D2000 OB2     10      -1.05         16   
+    11         ob      1    D2000 OB3     11      -1.05         16   
+    12         ob      1    D2000 OB4     12      -1.05         16   
+    13         ob      1    D2000 OB5     13      -1.05         16   
+    14         ob      1    D2000 OB6     14      -1.05         16   
+    15         ob      1    D2000 OB7     15      -1.05         16   
+    16         ob      1    D2000 OB8     16      -1.05         16   
+    17         ob      1    D2000 OB9     17      -1.05         16   
+    18         ob      1    D2000 OB10    18      -1.05         16   
+    19         ob      1    D2000 OB11    19      -1.05         16   
+    20         ob      1    D2000 OB12    20      -1.05         16   
+    21         ob      1    D2000 OB13    21      -1.05         16   
+    22         ob      1    D2000 OB14    22      -1.05         16   
+    23         ob      1    D2000 OB15    23      -1.05         16   
+    24         ob      1    D2000 OB16    24      -1.05         16   
+    25         ob      1    D2000 OB17    25      -1.05         16   
+    26         ob      1    D2000 OB18    26      -1.05         16   
+    27         ob      1    D2000 OB19    27      -1.05         16   
+    28         ob      1    D2000 OB20    28      -1.05         16   
+    29         feo     1    D2000 FEO1    29      1.575     55.845   
+    30         feo     1    D2000 FEO2    30      1.575     55.845   
+    31         feo     1    D2000 FEO3    31      1.575     55.845   
+    32         feo     1    D2000 FEO4    32      1.575     55.845   
+    33         st      1    D2000 ST1     33        2.1      28.09   
+    34         st      1    D2000 ST2     34        2.1      28.09   
+    35         st      1    D2000 ST3     35        2.1      28.09   
+    36         st      1    D2000 ST4     36        2.1      28.09   
+    37         st      1    D2000 ST5     37        2.1      28.09   
+    38         st      1    D2000 ST6     38        2.1      28.09   
+    39         st      1    D2000 ST7     39        2.1      28.09   
+    40         st      1    D2000 ST8     40        2.1      28.09   
+
+
+[ bonds ]
+; i	j	funct	length	force.c.
+1	5     1     0.1    463532.808
+2	6     1     0.1    463532.808
+3	7     1     0.1    463532.808
+4	8     1     0.1    463532.808