src/gromacs/gmxpreprocess/pdb2top.cpp

   1 /*
   2  * This file is part of the GROMACS molecular simulation package.
   3  *
   4  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
   5  * Copyright (c) 2001-2004, The GROMACS development team.
   6  * Copyright (c) 2013,2014,2015,2016,2017 by the GROMACS development team.
   7  * Copyright (c) 2018,2019,2020, by the GROMACS development team, led by
   8  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
   9  * and including many others, as listed in the AUTHORS file in the
  10  * top-level source directory and at http://www.gromacs.org.
  11  *
  12  * GROMACS is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU Lesser General Public License
  14  * as published by the Free Software Foundation; either version 2.1
  15  * of the License, or (at your option) any later version.
  16  *
  17  * GROMACS is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  20  * Lesser General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU Lesser General Public
  23  * License along with GROMACS; if not, see
  24  * http://www.gnu.org/licenses, or write to the Free Software Foundation,
  25  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA.
  26  *
  27  * If you want to redistribute modifications to GROMACS, please
  28  * consider that scientific software is very special. Version
  29  * control is crucial - bugs must be traceable. We will be happy to
  30  * consider code for inclusion in the official distribution, but
  31  * derived work must not be called official GROMACS. Details are found
  32  * in the README & COPYING files - if they are missing, get the
  33  * official version at http://www.gromacs.org.
  34  *
  35  * To help us fund GROMACS development, we humbly ask that you cite
  36  * the research papers on the package. Check out http://www.gromacs.org.
  37  */
  38 #include "gmxpre.h"
  39
  40 #include "pdb2top.h"
  41
  42 #include <cctype>
  43 #include <cmath>
  44 #include <cstdio>
  45 #include <cstring>
  46
  47 #include <algorithm>
  48 #include <string>
  49 #include <vector>
  50
  51 #include "gromacs/fileio/pdbio.h"
  52 #include "gromacs/gmxpreprocess/add_par.h"
  53 #include "gromacs/gmxpreprocess/fflibutil.h"
  54 #include "gromacs/gmxpreprocess/gen_ad.h"
  55 #include "gromacs/gmxpreprocess/gen_vsite.h"
  56 #include "gromacs/gmxpreprocess/grompp_impl.h"
  57 #include "gromacs/gmxpreprocess/h_db.h"
  58 #include "gromacs/gmxpreprocess/notset.h"
  59 #include "gromacs/gmxpreprocess/pgutil.h"
  60 #include "gromacs/gmxpreprocess/specbond.h"
  61 #include "gromacs/gmxpreprocess/topdirs.h"
  62 #include "gromacs/gmxpreprocess/topio.h"
  63 #include "gromacs/gmxpreprocess/toputil.h"
  64 #include "gromacs/math/functions.h"
  65 #include "gromacs/math/vec.h"
  66 #include "gromacs/topology/residuetypes.h"
  67 #include "gromacs/topology/symtab.h"
  68 #include "gromacs/utility/binaryinformation.h"
  69 #include "gromacs/utility/cstringutil.h"
  70 #include "gromacs/utility/dir_separator.h"
  71 #include "gromacs/utility/exceptions.h"
  72 #include "gromacs/utility/fatalerror.h"
  73 #include "gromacs/utility/futil.h"
  74 #include "gromacs/utility/logger.h"
  75 #include "gromacs/utility/niceheader.h"
  76 #include "gromacs/utility/path.h"
  77 #include "gromacs/utility/programcontext.h"
  78 #include "gromacs/utility/smalloc.h"
  79 #include "gromacs/utility/strconvert.h"
  80 #include "gromacs/utility/strdb.h"
  81 #include "gromacs/utility/stringutil.h"
  82 #include "gromacs/utility/textwriter.h"
  83
  84 #include "hackblock.h"
  85 #include "resall.h"
  86
  87 /* this must correspond to enum in pdb2top.h */
  88 const char* hh[ehisNR] = { "HISD", "HISE", "HISH", "HIS1" };
  89
  90 static int missing_atoms(const PreprocessResidue* rp, int resind, t_atoms* at, int i0, int i, const gmx::MDLogger& logger)
  91 {
  92     int nmiss = 0;
  93     for (int j = 0; j < rp->natom(); j++)
  94     {
  95         const char* name   = *(rp->atomname[j]);
  96         bool        bFound = false;
  97         for (int k = i0; k < i; k++)
  98         {
  99             bFound = (bFound || (gmx_strcasecmp(*(at->atomname[k]), name) == 0));
 100         }
 101         if (!bFound)
 102         {
 103             nmiss++;
 104             GMX_LOG(logger.warning)
 105                     .asParagraph()
 106                     .appendTextFormatted("atom %s is missing in residue %s %d in the pdb file",
 107                                          name,
 108                                          *(at->resinfo[resind].name),
 109                                          at->resinfo[resind].nr);
 110             if (name[0] == 'H' || name[0] == 'h')
 111             {
 112                 GMX_LOG(logger.warning)
 113                         .asParagraph()
 114                         .appendTextFormatted(
 115                                 "You might need to add atom %s to the hydrogen database of "
 116                                 "building block %s "
 117                                 "in the file %s.hdb (see the manual)",
 118                                 name,
 119                                 *(at->resinfo[resind].rtp),
 120                                 rp->filebase.c_str());
 121             }
 122         }
 123     }
 124
 125     return nmiss;
 126 }
 127
 128 bool is_int(double x)
 129 {
 130     const double tol = 1e-4;
 131     int          ix;
 132
 133     if (x < 0)
 134     {
 135         x = -x;
 136     }
 137     ix = gmx::roundToInt(x);
 138
 139     return (fabs(x - ix) < tol);
 140 }
 141
 142 static void choose_ff_impl(const char*          ffsel,
 143                            char*                forcefield,
 144                            int                  ff_maxlen,
 145                            char*                ffdir,
 146                            int                  ffdir_maxlen,
 147                            const gmx::MDLogger& logger)
 148 {
 149     std::vector<gmx::DataFileInfo> ffdirs = fflib_enumerate_forcefields();
 150     const int                      nff    = ssize(ffdirs);
 151
 152     /* Replace with unix path separators */
 153 #if DIR_SEPARATOR != '/'
 154     for (int i = 0; i < nff; ++i)
 155     {
 156         std::replace(ffdirs[i].dir.begin(), ffdirs[i].dir.end(), DIR_SEPARATOR, '/');
 157     }
 158 #endif
 159
 160     /* Store the force field names in ffs */
 161     std::vector<std::string> ffs;
 162     ffs.reserve(ffdirs.size());
 163     for (int i = 0; i < nff; ++i)
 164     {
 165         ffs.push_back(gmx::stripSuffixIfPresent(ffdirs[i].name, fflib_forcefield_dir_ext()));
 166     }
 167
 168     int sel;
 169     if (ffsel != nullptr)
 170     {
 171         sel        = -1;
 172         int cwdsel = -1;
 173         int nfound = 0;
 174         for (int i = 0; i < nff; ++i)
 175         {
 176             if (ffs[i] == ffsel)
 177             {
 178                 /* Matching ff name */
 179                 sel = i;
 180                 nfound++;
 181
 182                 if (ffdirs[i].dir == ".")
 183                 {
 184                     cwdsel = i;
 185                 }
 186             }
 187         }
 188
 189         if (cwdsel != -1)
 190         {
 191             sel = cwdsel;
 192         }
 193
 194         if (nfound > 1)
 195         {
 196             if (cwdsel != -1)
 197             {
 198                 GMX_LOG(logger.warning)
 199                         .asParagraph()
 200                         .appendTextFormatted(
 201                                 "Force field '%s' occurs in %d places. pdb2gmx is using the one in "
 202                                 "the current directory. Use interactive selection "
 203                                 "(not the -ff option) if you would prefer a different one.",
 204                                 ffsel,
 205                                 nfound);
 206             }
 207             else
 208             {
 209                 std::string message = gmx::formatString(
 210                         "Force field '%s' occurs in %d places, but not in "
 211                         "the current directory.\n"
 212                         "Run without the -ff switch and select the force "
 213                         "field interactively.",
 214                         ffsel,
 215                         nfound);
 216                 GMX_THROW(gmx::InconsistentInputError(message));
 217             }
 218         }
 219         else if (nfound == 0)
 220         {
 221             std::string message = gmx::formatString(
 222                     "Could not find force field '%s' in current directory, "
 223                     "install tree or GMXLIB path.",
 224                     ffsel);
 225             GMX_THROW(gmx::InconsistentInputError(message));
 226         }
 227     }
 228     else if (nff > 1)
 229     {
 230         std::vector<std::string> desc;
 231         desc.reserve(ffdirs.size());
 232         for (int i = 0; i < nff; ++i)
 233         {
 234             std::string docFileName(gmx::Path::join(ffdirs[i].dir, ffdirs[i].name, fflib_forcefield_doc()));
 235             // TODO: Just try to open the file with a method that does not
 236             // throw/bail out with a fatal error instead of multiple checks.
 237             if (gmx::File::exists(docFileName, gmx::File::returnFalseOnError))
 238             {
 239                 // TODO: Use a C++ API without such an intermediate/fixed-length buffer.
 240                 char buf[STRLEN];
 241                 /* We don't use fflib_open, because we don't want printf's */
 242                 FILE* fp = gmx_ffopen(docFileName, "r");
 243                 get_a_line(fp, buf, STRLEN);
 244                 gmx_ffclose(fp);
 245                 desc.emplace_back(buf);
 246             }
 247             else
 248             {
 249                 desc.push_back(ffs[i]);
 250             }
 251         }
 252         /* Order force fields from the same dir alphabetically
 253          * and put deprecated force fields at the end.
 254          */
 255         for (int i = 0; i < nff; ++i)
 256         {
 257             for (int j = i + 1; j < nff; ++j)
 258             {
 259                 if (ffdirs[i].dir == ffdirs[j].dir
 260                     && ((desc[i][0] == '[' && desc[j][0] != '[')
 261                         || ((desc[i][0] == '[' || desc[j][0] != '[')
 262                             && gmx_strcasecmp(desc[i].c_str(), desc[j].c_str()) > 0)))
 263                 {
 264                     std::swap(ffdirs[i].name, ffdirs[j].name);
 265                     std::swap(ffs[i], ffs[j]);
 266                     std::swap(desc[i], desc[j]);
 267                 }
 268             }
 269         }
 270
 271         GMX_LOG(logger.info).asParagraph().appendTextFormatted("Select the Force Field:");
 272         for (int i = 0; i < nff; ++i)
 273         {
 274             if (i == 0 || ffdirs[i - 1].dir != ffdirs[i].dir)
 275             {
 276                 if (ffdirs[i].dir == ".")
 277                 {
 278                     GMX_LOG(logger.info)
 279                             .asParagraph()
 280                             .appendTextFormatted("From current directory:");
 281                 }
 282                 else
 283                 {
 284                     GMX_LOG(logger.info)
 285                             .asParagraph()
 286                             .appendTextFormatted("From '%s':", ffdirs[i].dir.c_str());
 287                 }
 288             }
 289             GMX_LOG(logger.info).asParagraph().appendTextFormatted("%2d: %s", i + 1, desc[i].c_str());
 290         }
 291
 292         sel = -1;
 293         // TODO: Add a C++ API for this.
 294         char  buf[STRLEN];
 295         char* pret;
 296         do
 297         {
 298             pret = fgets(buf, STRLEN, stdin);
 299
 300             if (pret != nullptr)
 301             {
 302                 sel = strtol(buf, nullptr, 10);
 303                 sel--;
 304             }
 305         } while (pret == nullptr || (sel < 0) || (sel >= nff));
 306
 307         /* Check for a current limitation of the fflib code.
 308          * It will always read from the first ff directory in the list.
 309          * This check assumes that the order of ffs matches the order
 310          * in which fflib_open searches ff library files.
 311          */
 312         for (int i = 0; i < sel; i++)
 313         {
 314             if (ffs[i] == ffs[sel])
 315             {
 316                 std::string message = gmx::formatString(
 317                         "Can only select the first of multiple force "
 318                         "field entries with directory name '%s%s' in "
 319                         "the list. If you want to use the next entry, "
 320                         "run pdb2gmx in a different directory, set GMXLIB "
 321                         "to point to the desired force field first, and/or "
 322                         "rename or move the force field directory present "
 323                         "in the current working directory.",
 324                         ffs[sel].c_str(),
 325                         fflib_forcefield_dir_ext());
 326                 GMX_THROW(gmx::NotImplementedError(message));
 327             }
 328         }
 329     }
 330     else
 331     {
 332         sel = 0;
 333     }
 334
 335     if (ffs[sel].length() >= static_cast<size_t>(ff_maxlen))
 336     {
 337         std::string message = gmx::formatString("Length of force field name (%d) >= maxlen (%d)",
 338                                                 static_cast<int>(ffs[sel].length()),
 339                                                 ff_maxlen);
 340         GMX_THROW(gmx::InvalidInputError(message));
 341     }
 342     strcpy(forcefield, ffs[sel].c_str());
 343
 344     std::string ffpath;
 345     if (ffdirs[sel].bFromDefaultDir)
 346     {
 347         ffpath = ffdirs[sel].name;
 348     }
 349     else
 350     {
 351         ffpath = gmx::Path::join(ffdirs[sel].dir, ffdirs[sel].name);
 352     }
 353     if (ffpath.length() >= static_cast<size_t>(ffdir_maxlen))
 354     {
 355         std::string message = gmx::formatString("Length of force field dir (%d) >= maxlen (%d)",
 356                                                 static_cast<int>(ffpath.length()),
 357                                                 ffdir_maxlen);
 358         GMX_THROW(gmx::InvalidInputError(message));
 359     }
 360     strcpy(ffdir, ffpath.c_str());
 361 }
 362
 363 void choose_ff(const char* ffsel, char* forcefield, int ff_maxlen, char* ffdir, int ffdir_maxlen, const gmx::MDLogger& logger)
 364 {
 365     try
 366     {
 367         choose_ff_impl(ffsel, forcefield, ff_maxlen, ffdir, ffdir_maxlen, logger);
 368     }
 369     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
 370 }
 371
 372 void choose_watermodel(const char* wmsel, const char* ffdir, char** watermodel, const gmx::MDLogger& logger)
 373 {
 374     const char* fn_watermodels = "watermodels.dat";
 375     FILE*       fp;
 376     char        buf[STRLEN];
 377     int         nwm, sel, i;
 378     char**      model;
 379     char*       pret;
 380
 381     if (strcmp(wmsel, "none") == 0)
 382     {
 383         *watermodel = nullptr;
 384
 385         return;
 386     }
 387     else if (strcmp(wmsel, "select") != 0)
 388     {
 389         *watermodel = gmx_strdup(wmsel);
 390
 391         return;
 392     }
 393
 394     std::string filename = gmx::Path::join(ffdir, fn_watermodels);
 395     if (!fflib_fexist(filename))
 396     {
 397         GMX_LOG(logger.warning)
 398                 .asParagraph()
 399                 .appendTextFormatted("No file '%s' found, will not include a water model", fn_watermodels);
 400         *watermodel = nullptr;
 401
 402         return;
 403     }
 404
 405     fp = fflib_open(filename);
 406     GMX_LOG(logger.info).asParagraph().appendTextFormatted("Select the Water Model:");
 407     nwm   = 0;
 408     model = nullptr;
 409     while (get_a_line(fp, buf, STRLEN))
 410     {
 411         srenew(model, nwm + 1);
 412         snew(model[nwm], STRLEN);
 413         sscanf(buf, "%s%n", model[nwm], &i);
 414         if (i > 0)
 415         {
 416             ltrim(buf + i);
 417             GMX_LOG(logger.info).asParagraph().appendTextFormatted("%2d: %s", nwm + 1, buf + i);
 418             nwm++;
 419         }
 420         else
 421         {
 422             sfree(model[nwm]);
 423         }
 424     }
 425     gmx_ffclose(fp);
 426     GMX_LOG(logger.info).asParagraph().appendTextFormatted("%2d: %s", nwm + 1, "None");
 427
 428     sel = -1;
 429     do
 430     {
 431         pret = fgets(buf, STRLEN, stdin);
 432
 433         if (pret != nullptr)
 434         {
 435             sel = strtol(buf, nullptr, 10);
 436             sel--;
 437         }
 438     } while (pret == nullptr || sel < 0 || sel > nwm);
 439
 440     if (sel == nwm)
 441     {
 442         *watermodel = nullptr;
 443     }
 444     else
 445     {
 446         *watermodel = gmx_strdup(model[sel]);
 447     }
 448
 449     for (i = 0; i < nwm; i++)
 450     {
 451         sfree(model[i]);
 452     }
 453     sfree(model);
 454 }
 455
 456 static int name2type(t_atoms*                               at,
 457                      int**                                  cgnr,
 458                      gmx::ArrayRef<const PreprocessResidue> usedPpResidues,
 459                      ResidueType*                           rt,
 460                      const gmx::MDLogger&                   logger)
 461 {
 462     int    i, j, prevresind, i0, prevcg, cg, curcg;
 463     char*  name;
 464     bool   bNterm;
 465     double qt;
 466     int    nmissat;
 467
 468     nmissat = 0;
 469
 470     int resind = -1;
 471     bNterm     = false;
 472     i0         = 0;
 473     snew(*cgnr, at->nr);
 474     qt    = 0;
 475     curcg = 0;
 476     cg    = -1;
 477
 478     for (i = 0; (i < at->nr); i++)
 479     {
 480         prevresind = resind;
 481         if (at->atom[i].resind != resind)
 482         {
 483             resind     = at->atom[i].resind;
 484             bool bProt = rt->namedResidueHasType(*(at->resinfo[resind].name), "Protein");
 485             bNterm     = bProt && (resind == 0);
 486             if (resind > 0)
 487             {
 488                 nmissat += missing_atoms(&usedPpResidues[prevresind], prevresind, at, i0, i, logger);
 489             }
 490             i0 = i;
 491         }
 492         if (at->atom[i].m == 0)
 493         {
 494             qt               = 0;
 495             prevcg           = cg;
 496             name             = *(at->atomname[i]);
 497             j                = search_jtype(usedPpResidues[resind], name, bNterm);
 498             at->atom[i].type = usedPpResidues[resind].atom[j].type;
 499             at->atom[i].q    = usedPpResidues[resind].atom[j].q;
 500             at->atom[i].m    = usedPpResidues[resind].atom[j].m;
 501             cg               = usedPpResidues[resind].cgnr[j];
 502             /* A charge group number -1 signals a separate charge group
 503              * for this atom.
 504              */
 505             if ((cg == -1) || (cg != prevcg) || (resind != prevresind))
 506             {
 507                 curcg++;
 508             }
 509         }
 510         else
 511         {
 512             cg = -1;
 513             if (is_int(qt))
 514             {
 515                 qt = 0;
 516                 curcg++;
 517             }
 518             qt += at->atom[i].q;
 519         }
 520         (*cgnr)[i]        = curcg;
 521         at->atom[i].typeB = at->atom[i].type;
 522         at->atom[i].qB    = at->atom[i].q;
 523         at->atom[i].mB    = at->atom[i].m;
 524     }
 525     nmissat += missing_atoms(&usedPpResidues[resind], resind, at, i0, i, logger);
 526
 527     return nmissat;
 528 }
 529
 530 static void print_top_heavy_H(FILE* out, real mHmult)
 531 {
 532     if (mHmult == 2.0)
 533     {
 534         fprintf(out, "; Using deuterium instead of hydrogen\n\n");
 535     }
 536     else if (mHmult == 4.0)
 537     {
 538         fprintf(out, "#define HEAVY_H\n\n");
 539     }
 540     else if (mHmult != 1.0)
 541     {
 542         fprintf(stderr,
 543                 "WARNING: unsupported proton mass multiplier (%g) "
 544                 "in pdb2top\n",
 545                 mHmult);
 546     }
 547 }
 548
 549 void print_top_comment(FILE* out, const char* filename, const char* ffdir, bool bITP)
 550 {
 551     char  ffdir_parent[STRLEN];
 552     char* p;
 553
 554     try
 555     {
 556         gmx::TextWriter writer(out);
 557         gmx::niceHeader(&writer, filename, ';');
 558         writer.writeLine(gmx::formatString(";\tThis is a %s topology file", bITP ? "include" : "standalone"));
 559         writer.writeLine(";");
 560
 561         gmx::BinaryInformationSettings settings;
 562         settings.generatedByHeader(true);
 563         settings.linePrefix(";\t");
 564         gmx::printBinaryInformation(&writer, gmx::getProgramContext(), settings);
 565     }
 566     GMX_CATCH_ALL_AND_EXIT_WITH_FATAL_ERROR
 567
 568     if (strchr(ffdir, '/') == nullptr)
 569     {
 570         fprintf(out, ";\tForce field was read from the standard GROMACS share directory.\n;\n\n");
 571     }
 572     else if (ffdir[0] == '.')
 573     {
 574         fprintf(out,
 575                 ";\tForce field was read from current directory or a relative path - path "
 576                 "added.\n;\n\n");
 577     }
 578     else
 579     {
 580         strncpy(ffdir_parent, ffdir, STRLEN - 1);
 581         ffdir_parent[STRLEN - 1] = '\0'; /*make sure it is 0-terminated even for long string*/
 582         p                        = strrchr(ffdir_parent, '/');
 583
 584         *p = '\0';
 585
 586         fprintf(out,
 587                 ";\tForce field data was read from:\n"
 588                 ";\t%s\n"
 589                 ";\n"
 590                 ";\tNote:\n"
 591                 ";\tThis might be a non-standard force field location. When you use this topology, "
 592                 "the\n"
 593                 ";\tforce field must either be present in the current directory, or the location\n"
 594                 ";\tspecified in the GMXLIB path variable or with the 'include' mdp file "
 595                 "option.\n;\n\n",
 596                 ffdir_parent);
 597     }
 598 }
 599
 600 void print_top_header(FILE* out, const char* filename, bool bITP, const char* ffdir, real mHmult)
 601 {
 602     const char* p;
 603
 604     print_top_comment(out, filename, ffdir, bITP);
 605
 606     print_top_heavy_H(out, mHmult);
 607     fprintf(out, "; Include forcefield parameters\n");
 608
 609     p = strrchr(ffdir, '/');
 610     p = (ffdir[0] == '.' || p == nullptr) ? ffdir : p + 1;
 611
 612     fprintf(out, "#include \"%s/%s\"\n\n", p, fflib_forcefield_itp());
 613 }
 614
 615 static void print_top_posre(FILE* out, const char* pr)
 616 {
 617     fprintf(out, "; Include Position restraint file\n");
 618     fprintf(out, "#ifdef POSRES\n");
 619     fprintf(out, "#include \"%s\"\n", pr);
 620     fprintf(out, "#endif\n\n");
 621 }
 622
 623 static void print_top_water(FILE* out, const char* ffdir, const char* water)
 624 {
 625     const char* p;
 626     char        buf[STRLEN];
 627
 628     fprintf(out, "; Include water topology\n");
 629
 630     p = strrchr(ffdir, '/');
 631     p = (ffdir[0] == '.' || p == nullptr) ? ffdir : p + 1;
 632     fprintf(out, "#include \"%s/%s.itp\"\n", p, water);
 633
 634     fprintf(out, "\n");
 635     fprintf(out, "#ifdef POSRES_WATER\n");
 636     fprintf(out, "; Position restraint for each water oxygen\n");
 637     fprintf(out, "[ position_restraints ]\n");
 638     fprintf(out, ";%3s %5s %9s %10s %10s\n", "i", "funct", "fcx", "fcy", "fcz");
 639     fprintf(out, "%4d %4d %10g %10g %10g\n", 1, 1, 1000.0, 1000.0, 1000.0);
 640     fprintf(out, "#endif\n");
 641     fprintf(out, "\n");
 642
 643     sprintf(buf, "%s/ions.itp", p);
 644
 645     if (fflib_fexist(buf))
 646     {
 647         fprintf(out, "; Include topology for ions\n");
 648         fprintf(out, "#include \"%s\"\n", buf);
 649         fprintf(out, "\n");
 650     }
 651 }
 652
 653 static void print_top_system(FILE* out, const char* title)
 654 {
 655     fprintf(out, "[ %s ]\n", dir2str(Directive::d_system));
 656     fprintf(out, "; Name\n");
 657     fprintf(out, "%s\n\n", title[0] ? title : "Protein");
 658 }
 659
 660 void print_top_mols(FILE*                            out,
 661                     const char*                      title,
 662                     const char*                      ffdir,
 663                     const char*                      water,
 664                     gmx::ArrayRef<const std::string> incls,
 665                     gmx::ArrayRef<const t_mols>      mols)
 666 {
 667
 668     if (!incls.empty())
 669     {
 670         fprintf(out, "; Include chain topologies\n");
 671         for (const auto& incl : incls)
 672         {
 673             fprintf(out, "#include \"%s\"\n", gmx::Path::getFilename(incl).c_str());
 674         }
 675         fprintf(out, "\n");
 676     }
 677
 678     if (water)
 679     {
 680         print_top_water(out, ffdir, water);
 681     }
 682     print_top_system(out, title);
 683
 684     if (!mols.empty())
 685     {
 686         fprintf(out, "[ %s ]\n", dir2str(Directive::d_molecules));
 687         fprintf(out, "; %-15s %5s\n", "Compound", "#mols");
 688         for (const auto& mol : mols)
 689         {
 690             fprintf(out, "%-15s %5d\n", mol.name.c_str(), mol.nr);
 691         }
 692     }
 693 }
 694
 695 void write_top(FILE*                                   out,
 696                const char*                             pr,
 697                const char*                             molname,
 698                t_atoms*                                at,
 699                bool                                    bRTPresname,
 700                int                                     bts[],
 701                gmx::ArrayRef<const InteractionsOfType> plist,
 702                t_excls                                 excls[],
 703                PreprocessingAtomTypes*                 atype,
 704                int*                                    cgnr,
 705                int                                     nrexcl)
 706 /* NOTE: nrexcl is not the size of *excl! */
 707 {
 708     if (at && atype && cgnr)
 709     {
 710         fprintf(out, "[ %s ]\n", dir2str(Directive::d_moleculetype));
 711         fprintf(out, "; %-15s %5s\n", "Name", "nrexcl");
 712         fprintf(out, "%-15s %5d\n\n", molname ? molname : "Protein", nrexcl);
 713
 714         print_atoms(out, atype, at, cgnr, bRTPresname);
 715         print_bondeds(out, at->nr, Directive::d_bonds, F_BONDS, bts[ebtsBONDS], plist);
 716         print_bondeds(out, at->nr, Directive::d_constraints, F_CONSTR, 0, plist);
 717         print_bondeds(out, at->nr, Directive::d_constraints, F_CONSTRNC, 0, plist);
 718         print_bondeds(out, at->nr, Directive::d_pairs, F_LJ14, 0, plist);
 719         print_excl(out, at->nr, excls);
 720         print_bondeds(out, at->nr, Directive::d_angles, F_ANGLES, bts[ebtsANGLES], plist);
 721         print_bondeds(out, at->nr, Directive::d_dihedrals, F_PDIHS, bts[ebtsPDIHS], plist);
 722         print_bondeds(out, at->nr, Directive::d_dihedrals, F_IDIHS, bts[ebtsIDIHS], plist);
 723         print_bondeds(out, at->nr, Directive::d_cmap, F_CMAP, bts[ebtsCMAP], plist);
 724         print_bondeds(out, at->nr, Directive::d_polarization, F_POLARIZATION, 0, plist);
 725         print_bondeds(out, at->nr, Directive::d_thole_polarization, F_THOLE_POL, 0, plist);
 726         print_bondeds(out, at->nr, Directive::d_vsites2, F_VSITE2, 0, plist);
 727         print_bondeds(out, at->nr, Directive::d_vsites3, F_VSITE3, 0, plist);
 728         print_bondeds(out, at->nr, Directive::d_vsites3, F_VSITE3FD, 0, plist);
 729         print_bondeds(out, at->nr, Directive::d_vsites3, F_VSITE3FAD, 0, plist);
 730         print_bondeds(out, at->nr, Directive::d_vsites3, F_VSITE3OUT, 0, plist);
 731         print_bondeds(out, at->nr, Directive::d_vsites4, F_VSITE4FD, 0, plist);
 732         print_bondeds(out, at->nr, Directive::d_vsites4, F_VSITE4FDN, 0, plist);
 733
 734         if (pr)
 735         {
 736             print_top_posre(out, pr);
 737         }
 738     }
 739 }
 740
 741
 742 static void do_ssbonds(InteractionsOfType*                ps,
 743                        t_atoms*                           atoms,
 744                        gmx::ArrayRef<const DisulfideBond> ssbonds,
 745                        bool                               bAllowMissing)
 746 {
 747     for (const auto& bond : ssbonds)
 748     {
 749         int ri = bond.firstResidue;
 750         int rj = bond.secondResidue;
 751         int ai = search_res_atom(bond.firstAtom.c_str(), ri, atoms, "special bond", bAllowMissing);
 752         int aj = search_res_atom(bond.secondAtom.c_str(), rj, atoms, "special bond", bAllowMissing);
 753         if ((ai == -1) || (aj == -1))
 754         {
 755             gmx_fatal(FARGS,
 756                       "Trying to make impossible special bond (%s-%s)!",
 757                       bond.firstAtom.c_str(),
 758                       bond.secondAtom.c_str());
 759         }
 760         add_param(ps, ai, aj, {}, nullptr);
 761     }
 762 }
 763
 764 static void at2bonds(InteractionsOfType*                  psb,
 765                      gmx::ArrayRef<MoleculePatchDatabase> globalPatches,
 766                      t_atoms*                             atoms,
 767                      gmx::ArrayRef<const gmx::RVec>       x,
 768                      real                                 long_bond_dist,
 769                      real                                 short_bond_dist,
 770                      gmx::ArrayRef<const int>             cyclicBondsIndex,
 771                      const gmx::MDLogger&                 logger)
 772 {
 773     real        long_bond_dist2, short_bond_dist2;
 774     const char* ptr;
 775
 776     long_bond_dist2  = gmx::square(long_bond_dist);
 777     short_bond_dist2 = gmx::square(short_bond_dist);
 778
 779     if (debug)
 780     {
 781         ptr = "bond";
 782     }
 783     else
 784     {
 785         ptr = "check";
 786     }
 787
 788     GMX_LOG(logger.info).asParagraph().appendTextFormatted("Making bonds...");
 789     int i = 0;
 790     for (int resind = 0; (resind < atoms->nres) && (i < atoms->nr); resind++)
 791     {
 792         /* add bonds from list of bonded interactions */
 793         for (const auto& patch : globalPatches[resind].rb[ebtsBONDS].b)
 794         {
 795             /* Unfortunately we can not issue errors or warnings
 796              * for missing atoms in bonds, as the hydrogens and terminal atoms
 797              * have not been added yet.
 798              */
 799             int ai = search_atom(patch.ai().c_str(), i, atoms, ptr, TRUE, cyclicBondsIndex);
 800             int aj = search_atom(patch.aj().c_str(), i, atoms, ptr, TRUE, cyclicBondsIndex);
 801             if (ai != -1 && aj != -1)
 802             {
 803                 real dist2 = distance2(x[ai], x[aj]);
 804                 if (dist2 > long_bond_dist2)
 805
 806                 {
 807                     GMX_LOG(logger.warning)
 808                             .asParagraph()
 809                             .appendTextFormatted(
 810                                     "Long Bond (%d-%d = %g nm)", ai + 1, aj + 1, std::sqrt(dist2));
 811                 }
 812                 else if (dist2 < short_bond_dist2)
 813                 {
 814                     GMX_LOG(logger.warning)
 815                             .asParagraph()
 816                             .appendTextFormatted(
 817                                     "Short Bond (%d-%d = %g nm)", ai + 1, aj + 1, std::sqrt(dist2));
 818                 }
 819                 add_param(psb, ai, aj, {}, patch.s.c_str());
 820             }
 821         }
 822         /* add bonds from list of hacks (each added atom gets a bond) */
 823         while ((i < atoms->nr) && (atoms->atom[i].resind == resind))
 824         {
 825             for (const auto& patch : globalPatches[resind].hack)
 826             {
 827                 if ((patch.tp > 0 || patch.type() == MoleculePatchType::Add)
 828                     && patch.a[0] == *(atoms->atomname[i]))
 829                 {
 830                     switch (patch.tp)
 831                     {
 832                         case 9:                                        /* COOH terminus */
 833                             add_param(psb, i, i + 1, {}, nullptr);     /* C-O  */
 834                             add_param(psb, i, i + 2, {}, nullptr);     /* C-OA */
 835                             add_param(psb, i + 2, i + 3, {}, nullptr); /* OA-H */
 836                             break;
 837                         default:
 838                             for (int k = 0; (k < patch.nr); k++)
 839                             {
 840                                 add_param(psb, i, i + k + 1, {}, nullptr);
 841                             }
 842                     }
 843                 }
 844             }
 845             i++;
 846         }
 847         /* we're now at the start of the next residue */
 848     }
 849 }
 850
 851 static bool pcompar(const InteractionOfType& a, const InteractionOfType& b)
 852 {
 853     int d;
 854
 855     if (((d = a.ai() - b.ai()) != 0) || ((d = a.aj() - b.aj()) != 0))
 856     {
 857         return d < 0;
 858     }
 859     else
 860     {
 861         return a.interactionTypeName().length() > b.interactionTypeName().length();
 862     }
 863 }
 864
 865 static void clean_bonds(InteractionsOfType* ps, const gmx::MDLogger& logger)
 866 {
 867     if (ps->size() > 0)
 868     {
 869         /* Sort bonds */
 870         for (auto& bond : ps->interactionTypes)
 871         {
 872             bond.sortAtomIds();
 873         }
 874         std::sort(ps->interactionTypes.begin(), ps->interactionTypes.end(), pcompar);
 875
 876         /* remove doubles, keep the first one always. */
 877         int oldNumber = ps->size();
 878         for (auto parm = ps->interactionTypes.begin() + 1; parm != ps->interactionTypes.end();)
 879         {
 880             auto prev = parm - 1;
 881             if (parm->ai() == prev->ai() && parm->aj() == prev->aj())
 882             {
 883                 parm = ps->interactionTypes.erase(parm);
 884             }
 885             else
 886             {
 887                 ++parm;
 888             }
 889         }
 890         GMX_LOG(logger.info)
 891                 .asParagraph()
 892                 .appendTextFormatted("Number of bonds was %d, now %zu", oldNumber, ps->size());
 893     }
 894     else
 895     {
 896         GMX_LOG(logger.info).asParagraph().appendTextFormatted("No bonds");
 897     }
 898 }
 899
 900 void print_sums(const t_atoms* atoms, bool bSystem, const gmx::MDLogger& logger)
 901 {
 902     double      m, qtot;
 903     int         i;
 904     const char* where;
 905
 906     if (bSystem)
 907     {
 908         where = " in system";
 909     }
 910     else
 911     {
 912         where = "";
 913     }
 914
 915     m    = 0;
 916     qtot = 0;
 917     for (i = 0; (i < atoms->nr); i++)
 918     {
 919         m += atoms->atom[i].m;
 920         qtot += atoms->atom[i].q;
 921     }
 922
 923     GMX_LOG(logger.info).asParagraph().appendTextFormatted("Total mass%s %.3f a.m.u.", where, m);
 924     GMX_LOG(logger.info).asParagraph().appendTextFormatted("Total charge%s %.3f e", where, qtot);
 925 }
 926
 927 static void check_restp_type(const char* name, int t1, int t2)
 928 {
 929     if (t1 != t2)
 930     {
 931         gmx_fatal(FARGS, "Residues in one molecule have a different '%s' type: %d and %d", name, t1, t2);
 932     }
 933 }
 934
 935 static void check_restp_types(const PreprocessResidue& r0, const PreprocessResidue& r1)
 936 {
 937     check_restp_type("all dihedrals",
 938                      static_cast<int>(r0.bKeepAllGeneratedDihedrals),
 939                      static_cast<int>(r1.bKeepAllGeneratedDihedrals));
 940     check_restp_type("nrexcl", r0.nrexcl, r1.nrexcl);
 941     check_restp_type("HH14",
 942                      static_cast<int>(r0.bGenerateHH14Interactions),
 943                      static_cast<int>(r1.bGenerateHH14Interactions));
 944     check_restp_type("remove dihedrals",
 945                      static_cast<int>(r0.bRemoveDihedralIfWithImproper),
 946                      static_cast<int>(r1.bRemoveDihedralIfWithImproper));
 947
 948     for (int i = 0; i < ebtsNR; i++)
 949     {
 950         check_restp_type(btsNames[i], r0.rb[i].type, r1.rb[i].type);
 951     }
 952 }
 953
 954 static void add_atom_to_restp(PreprocessResidue*   usedPpResidues,
 955                               t_symtab*            symtab,
 956                               int                  at_start,
 957                               const MoleculePatch* patch)
 958 {
 959     /* now add them */
 960     for (int k = 0; k < patch->nr; k++)
 961     {
 962         /* set counter in atomname */
 963         std::string buf = patch->nname;
 964         if (patch->nr > 1)
 965         {
 966             buf.append(gmx::formatString("%d", k + 1));
 967         }
 968         usedPpResidues->atomname.insert(usedPpResidues->atomname.begin() + at_start + 1 + k,
 969                                         put_symtab(symtab, buf.c_str()));
 970         usedPpResidues->atom.insert(usedPpResidues->atom.begin() + at_start + 1 + k, patch->atom.back());
 971         if (patch->cgnr != NOTSET)
 972         {
 973             usedPpResidues->cgnr.insert(usedPpResidues->cgnr.begin() + at_start + 1 + k, patch->cgnr);
 974         }
 975         else
 976         {
 977             usedPpResidues->cgnr.insert(usedPpResidues->cgnr.begin() + at_start + 1 + k,
 978                                         usedPpResidues->cgnr[at_start]);
 979         }
 980     }
 981 }
 982
 983 void get_hackblocks_rtp(std::vector<MoleculePatchDatabase>*    globalPatches,
 984                         std::vector<PreprocessResidue>*        usedPpResidues,
 985                         gmx::ArrayRef<const PreprocessResidue> rtpFFDB,
 986                         int                                    nres,
 987                         t_resinfo*                             resinfo,
 988                         int                                    nterpairs,
 989                         t_symtab*                              symtab,
 990                         gmx::ArrayRef<MoleculePatchDatabase*>  ntdb,
 991                         gmx::ArrayRef<MoleculePatchDatabase*>  ctdb,
 992                         gmx::ArrayRef<const int>               rn,
 993                         gmx::ArrayRef<const int>               rc,
 994                         bool                                   bAllowMissing,
 995                         const gmx::MDLogger&                   logger)
 996 {
 997     char* key;
 998     bool  bRM;
 999
1000     globalPatches->resize(nres);
1001     usedPpResidues->clear();
1002     /* first the termini */
1003     for (int i = 0; i < nterpairs; i++)
1004     {
1005         if (rn[i] >= 0 && ntdb[i] != nullptr)
1006         {
1007             copyModificationBlocks(*ntdb[i], &globalPatches->at(rn[i]));
1008         }
1009         if (rc[i] >= 0 && ctdb[i] != nullptr)
1010         {
1011             mergeAtomAndBondModifications(*ctdb[i], &globalPatches->at(rc[i]));
1012         }
1013     }
1014
1015     /* then the whole rtp */
1016     for (int i = 0; i < nres; i++)
1017     {
1018         /* Here we allow a mismatch of one character when looking for the rtp entry.
1019          * For such a mismatch there should be only one mismatching name.
1020          * This is mainly useful for small molecules such as ions.
1021          * Note that this will usually not work for protein, DNA and RNA,
1022          * since there the residue names should be listed in residuetypes.dat
1023          * and an error will have been generated earlier in the process.
1024          */
1025         key = *resinfo[i].rtp;
1026
1027         resinfo[i].rtp = put_symtab(symtab, searchResidueDatabase(key, rtpFFDB, logger).c_str());
1028         auto res       = getDatabaseEntry(*resinfo[i].rtp, rtpFFDB);
1029         usedPpResidues->push_back(PreprocessResidue());
1030         PreprocessResidue* newentry = &usedPpResidues->back();
1031         copyPreprocessResidues(*res, newentry, symtab);
1032
1033         /* Check that we do not have different bonded types in one molecule */
1034         check_restp_types(usedPpResidues->front(), *newentry);
1035
1036         int tern = -1;
1037         for (int j = 0; j < nterpairs && tern == -1; j++)
1038         {
1039             if (i == rn[j])
1040             {
1041                 tern = j;
1042             }
1043         }
1044         int terc = -1;
1045         for (int j = 0; j < nterpairs && terc == -1; j++)
1046         {
1047             if (i == rc[j])
1048             {
1049                 terc = j;
1050             }
1051         }
1052         bRM = mergeBondedInteractionList(res->rb, globalPatches->at(i).rb, tern >= 0, terc >= 0);
1053
1054         if (bRM && ((tern >= 0 && ntdb[tern] == nullptr) || (terc >= 0 && ctdb[terc] == nullptr)))
1055         {
1056             const char* errString =
1057                     "There is a dangling bond at at least one of the terminal ends and the force "
1058                     "field does not provide terminal entries or files. Fix your terminal "
1059                     "residues so that they match the residue database (.rtp) entries, or provide "
1060                     "terminal database entries (.tdb).";
1061             if (bAllowMissing)
1062             {
1063                 GMX_LOG(logger.warning).asParagraph().appendTextFormatted("%s", errString);
1064             }
1065             else
1066             {
1067                 gmx_fatal(FARGS, "%s", errString);
1068             }
1069         }
1070         else if (bRM && ((tern >= 0 && ntdb[tern]->nhack() == 0) || (terc >= 0 && ctdb[terc]->nhack() == 0)))
1071         {
1072             const char* errString =
1073                     "There is a dangling bond at at least one of the terminal ends. Fix your "
1074                     "coordinate file, add a new terminal database entry (.tdb), or select the "
1075                     "proper existing terminal entry.";
1076             if (bAllowMissing)
1077             {
1078                 GMX_LOG(logger.warning).asParagraph().appendTextFormatted("%s", errString);
1079             }
1080             else
1081             {
1082                 gmx_fatal(FARGS, "%s", errString);
1083             }
1084         }
1085     }
1086
1087     /* Apply patchs to t_restp entries
1088        i.e. add's and deletes from termini database will be
1089        added to/removed from residue topology
1090        we'll do this on one big dirty loop, so it won't make easy reading! */
1091     for (auto modifiedResidue = globalPatches->begin(); modifiedResidue != globalPatches->end();
1092          modifiedResidue++)
1093     {
1094         const int          pos    = std::distance(globalPatches->begin(), modifiedResidue);
1095         PreprocessResidue* posres = &usedPpResidues->at(pos);
1096         for (auto patch = modifiedResidue->hack.begin(); patch != modifiedResidue->hack.end(); patch++)
1097         {
1098             if (patch->nr != 0)
1099             {
1100                 /* find atom in restp */
1101                 auto found = std::find_if(
1102                         posres->atomname.begin(), posres->atomname.end(), [&patch](char** name) {
1103                             return (patch->oname.empty() && patch->a[0] == *name)
1104                                    || (patch->oname == *name);
1105                         });
1106
1107                 if (found == posres->atomname.end())
1108                 {
1109                     /* If we are doing an atom rename only, we don't need
1110                      * to generate a fatal error if the old name is not found
1111                      * in the rtp.
1112                      */
1113                     /* Deleting can happen also only on the input atoms,
1114                      * not necessarily always on the rtp entry.
1115                      */
1116                     if (patch->type() == MoleculePatchType::Add)
1117                     {
1118                         gmx_fatal(FARGS,
1119                                   "atom %s not found in buiding block %d%s "
1120                                   "while combining tdb and rtp",
1121                                   patch->oname.empty() ? patch->a[0].c_str() : patch->oname.c_str(),
1122                                   pos + 1,
1123                                   *resinfo[pos].rtp);
1124                     }
1125                 }
1126                 else
1127                 {
1128                     int l = std::distance(posres->atomname.begin(), found);
1129                     switch (patch->type())
1130                     {
1131                         case MoleculePatchType::Add:
1132                         {
1133                             /* we're adding: */
1134                             add_atom_to_restp(posres, symtab, l, &(*patch));
1135                             break;
1136                         }
1137                         case MoleculePatchType::Delete:
1138                         { /* we're deleting */
1139                             posres->atom.erase(posres->atom.begin() + l);
1140                             posres->atomname.erase(posres->atomname.begin() + l);
1141                             posres->cgnr.erase(posres->cgnr.begin() + l);
1142                             break;
1143                         }
1144                         case MoleculePatchType::Replace:
1145                         {
1146                             /* we're replacing */
1147                             posres->atom[l]     = patch->atom.back();
1148                             posres->atomname[l] = put_symtab(symtab, patch->nname.c_str());
1149                             if (patch->cgnr != NOTSET)
1150                             {
1151                                 posres->cgnr[l] = patch->cgnr;
1152                             }
1153                             break;
1154                         }
1155                     }
1156                 }
1157             }
1158         }
1159     }
1160 }
1161
1162 static bool atomname_cmp_nr(const char* anm, const MoleculePatch* patch, int* nr)
1163 {
1164
1165     if (patch->nr == 1)
1166     {
1167         *nr = 0;
1168
1169         return (gmx::equalCaseInsensitive(anm, patch->nname));
1170     }
1171     else
1172     {
1173         if (isdigit(anm[strlen(anm) - 1]))
1174         {
1175             *nr = anm[strlen(anm) - 1] - '0';
1176         }
1177         else
1178         {
1179             *nr = 0;
1180         }
1181         if (*nr <= 0 || *nr > patch->nr)
1182         {
1183             return FALSE;
1184         }
1185         else
1186         {
1187             return (strlen(anm) == patch->nname.length() + 1
1188                     && gmx_strncasecmp(anm, patch->nname.c_str(), patch->nname.length()) == 0);
1189         }
1190     }
1191 }
1192
1193 static bool match_atomnames_with_rtp_atom(t_atoms*                     pdba,
1194                                           gmx::ArrayRef<gmx::RVec>     x,
1195                                           t_symtab*                    symtab,
1196                                           int                          atind,
1197                                           PreprocessResidue*           localPpResidue,
1198                                           const MoleculePatchDatabase& singlePatch,
1199                                           bool                         bVerbose,
1200                                           const gmx::MDLogger&         logger)
1201 {
1202     int   resnr;
1203     char* oldnm;
1204     int   anmnr;
1205     bool  bDeleted;
1206
1207     oldnm = *pdba->atomname[atind];
1208     resnr = pdba->resinfo[pdba->atom[atind].resind].nr;
1209
1210     bDeleted = FALSE;
1211     for (auto patch = singlePatch.hack.begin(); patch != singlePatch.hack.end(); patch++)
1212     {
1213         if (patch->type() == MoleculePatchType::Replace && gmx::equalCaseInsensitive(oldnm, patch->oname))
1214         {
1215             /* This is a replace entry. */
1216             /* Check if we are not replacing a replaced atom. */
1217             bool bReplaceReplace = false;
1218             for (auto selfPatch = singlePatch.hack.begin(); selfPatch != singlePatch.hack.end(); selfPatch++)
1219             {
1220                 if (patch != selfPatch && selfPatch->type() == MoleculePatchType::Replace
1221                     && gmx::equalCaseInsensitive(selfPatch->nname, patch->oname))
1222                 {
1223                     /* The replace in patch replaces an atom that
1224                      * was already replaced in selfPatch, we do not want
1225                      * second or higher level replaces at this stage.
1226                      */
1227                     bReplaceReplace = true;
1228                 }
1229             }
1230             if (bReplaceReplace)
1231             {
1232                 /* Skip this replace. */
1233                 continue;
1234             }
1235
1236             /* This atom still has the old name, rename it */
1237             std::string newnm = patch->nname;
1238             auto        found = std::find_if(
1239                     localPpResidue->atomname.begin(),
1240                     localPpResidue->atomname.end(),
1241                     [&newnm](char** name) { return gmx::equalCaseInsensitive(newnm, *name); });
1242             if (found == localPpResidue->atomname.end())
1243             {
1244                 /* The new name is not present in the rtp.
1245                  * We need to apply the replace also to the rtp entry.
1246                  */
1247
1248                 /* We need to find the add hack that can add this atom
1249                  * to find out after which atom it should be added.
1250                  */
1251                 bool bFoundInAdd = false;
1252                 for (auto rtpModification = singlePatch.hack.begin();
1253                      rtpModification != singlePatch.hack.end();
1254                      rtpModification++)
1255                 {
1256                     int         k = std::distance(localPpResidue->atomname.begin(), found);
1257                     std::string start_at;
1258                     if (rtpModification->type() == MoleculePatchType::Add
1259                         && atomname_cmp_nr(newnm.c_str(), &(*rtpModification), &anmnr))
1260                     {
1261                         if (anmnr <= 1)
1262                         {
1263                             start_at = singlePatch.hack[k].a[0];
1264                         }
1265                         else
1266                         {
1267                             start_at = gmx::formatString(
1268                                     "%s%d", singlePatch.hack[k].nname.c_str(), anmnr - 1);
1269                         }
1270                         auto found2 = std::find_if(localPpResidue->atomname.begin(),
1271                                                    localPpResidue->atomname.end(),
1272                                                    [&start_at](char** name) {
1273                                                        return gmx::equalCaseInsensitive(start_at, *name);
1274                                                    });
1275                         if (found2 == localPpResidue->atomname.end())
1276                         {
1277                             gmx_fatal(FARGS,
1278                                       "Could not find atom '%s' in residue building block '%s' to "
1279                                       "add atom '%s' to",
1280                                       start_at.c_str(),
1281                                       localPpResidue->resname.c_str(),
1282                                       newnm.c_str());
1283                         }
1284                         /* We can add the atom after atom start_nr */
1285                         add_atom_to_restp(localPpResidue,
1286                                           symtab,
1287                                           std::distance(localPpResidue->atomname.begin(), found2),
1288                                           &(*patch));
1289
1290                         bFoundInAdd = true;
1291                     }
1292                 }
1293
1294                 if (!bFoundInAdd)
1295                 {
1296                     gmx_fatal(FARGS,
1297                               "Could not find an 'add' entry for atom named '%s' corresponding to "
1298                               "the 'replace' entry from atom name '%s' to '%s' for tdb or hdb "
1299                               "database of residue type '%s'",
1300                               newnm.c_str(),
1301                               patch->oname.c_str(),
1302                               patch->nname.c_str(),
1303                               localPpResidue->resname.c_str());
1304                 }
1305             }
1306
1307             if (bVerbose)
1308             {
1309                 GMX_LOG(logger.info)
1310                         .asParagraph()
1311                         .appendTextFormatted("Renaming atom '%s' in residue '%s' %d to '%s'",
1312                                              oldnm,
1313                                              localPpResidue->resname.c_str(),
1314                                              resnr,
1315                                              newnm.c_str());
1316             }
1317             /* Rename the atom in pdba */
1318             pdba->atomname[atind] = put_symtab(symtab, newnm.c_str());
1319         }
1320         else if (patch->type() == MoleculePatchType::Delete
1321                  && gmx::equalCaseInsensitive(oldnm, patch->oname))
1322         {
1323             /* This is a delete entry, check if this atom is present
1324              * in the rtp entry of this residue.
1325              */
1326             auto found3 = std::find_if(
1327                     localPpResidue->atomname.begin(),
1328                     localPpResidue->atomname.end(),
1329                     [&oldnm](char** name) { return gmx::equalCaseInsensitive(oldnm, *name); });
1330             if (found3 == localPpResidue->atomname.end())
1331             {
1332                 /* This atom is not present in the rtp entry,
1333                  * delete is now from the input pdba.
1334                  */
1335                 if (bVerbose)
1336                 {
1337                     GMX_LOG(logger.info)
1338                             .asParagraph()
1339                             .appendTextFormatted("Deleting atom '%s' in residue '%s' %d",
1340                                                  oldnm,
1341                                                  localPpResidue->resname.c_str(),
1342                                                  resnr);
1343                 }
1344                 /* We should free the atom name,
1345                  * but it might be used multiple times in the symtab.
1346                  * sfree(pdba->atomname[atind]);
1347                  */
1348                 for (int k = atind + 1; k < pdba->nr; k++)
1349                 {
1350                     pdba->atom[k - 1]     = pdba->atom[k];
1351                     pdba->atomname[k - 1] = pdba->atomname[k];
1352                     copy_rvec(x[k], x[k - 1]);
1353                 }
1354                 pdba->nr--;
1355                 bDeleted = true;
1356             }
1357         }
1358     }
1359
1360     return bDeleted;
1361 }
1362
1363 void match_atomnames_with_rtp(gmx::ArrayRef<PreprocessResidue>     usedPpResidues,
1364                               gmx::ArrayRef<MoleculePatchDatabase> globalPatches,
1365                               t_atoms*                             pdba,
1366                               t_symtab*                            symtab,
1367                               gmx::ArrayRef<gmx::RVec>             x,
1368                               bool                                 bVerbose,
1369                               const gmx::MDLogger&                 logger)
1370 {
1371     for (int i = 0; i < pdba->nr; i++)
1372     {
1373         const char*        oldnm          = *pdba->atomname[i];
1374         PreprocessResidue* localPpResidue = &usedPpResidues[pdba->atom[i].resind];
1375         auto               found          = std::find_if(
1376                 localPpResidue->atomname.begin(), localPpResidue->atomname.end(), [&oldnm](char** name) {
1377                     return gmx::equalCaseInsensitive(oldnm, *name);
1378                 });
1379         if (found == localPpResidue->atomname.end())
1380         {
1381             /* Not found yet, check if we have to rename this atom */
1382             if (match_atomnames_with_rtp_atom(
1383                         pdba, x, symtab, i, localPpResidue, globalPatches[pdba->atom[i].resind], bVerbose, logger))
1384             {
1385                 /* We deleted this atom, decrease the atom counter by 1. */
1386                 i--;
1387             }
1388         }
1389     }
1390 }
1391
1392 #define NUM_CMAP_ATOMS 5
1393 static void gen_cmap(InteractionsOfType*                    psb,
1394                      gmx::ArrayRef<const PreprocessResidue> usedPpResidues,
1395                      t_atoms*                               atoms,
1396                      gmx::ArrayRef<const int>               cyclicBondsIndex,
1397                      const gmx::MDLogger&                   logger)
1398 {
1399     int         residx;
1400     const char* ptr;
1401     t_resinfo*  resinfo = atoms->resinfo;
1402     int         nres    = atoms->nres;
1403     int         cmap_atomid[NUM_CMAP_ATOMS];
1404     int         cmap_chainnum = -1;
1405
1406     if (debug)
1407     {
1408         ptr = "cmap";
1409     }
1410     else
1411     {
1412         ptr = "check";
1413     }
1414
1415     GMX_LOG(logger.info).asParagraph().appendTextFormatted("Making cmap torsions...");
1416     int i = 0;
1417     /* Most cmap entries use the N atom from the next residue, so the last
1418      * residue should not have its CMAP entry in that case, but for things like
1419      * dipeptides we sometimes define a complete CMAP entry inside a residue,
1420      * and in this case we need to process everything through the last residue.
1421      */
1422     for (residx = 0; residx < nres; residx++)
1423     {
1424         /* Add CMAP terms from the list of CMAP interactions */
1425         for (const auto& b : usedPpResidues[residx].rb[ebtsCMAP].b)
1426         {
1427             bool bAddCMAP = true;
1428             /* Loop over atoms in a candidate CMAP interaction and
1429              * check that they exist, are from the same chain and are
1430              * from residues labelled as protein. */
1431             for (int k = 0; k < NUM_CMAP_ATOMS && bAddCMAP; k++)
1432             {
1433                 /* Assign the pointer to the name of the next reference atom.
1434                  * This can use -/+ labels to refer to previous/next residue.
1435                  */
1436                 const char* pname = b.a[k].c_str();
1437                 /* Skip this CMAP entry if it refers to residues before the
1438                  * first or after the last residue.
1439                  */
1440                 if ((cyclicBondsIndex.empty() && ((strchr(pname, '-') != nullptr) && (residx == 0)))
1441                     || ((strchr(pname, '+') != nullptr) && (residx == nres - 1)))
1442                 {
1443                     bAddCMAP = false;
1444                     break;
1445                 }
1446
1447                 cmap_atomid[k] = search_atom(pname, i, atoms, ptr, TRUE, cyclicBondsIndex);
1448                 bAddCMAP       = bAddCMAP && (cmap_atomid[k] != -1);
1449                 if (!bAddCMAP)
1450                 {
1451                     /* This break is necessary, because cmap_atomid[k]
1452                      * == -1 cannot be safely used as an index
1453                      * into the atom array. */
1454                     break;
1455                 }
1456                 int this_residue_index = atoms->atom[cmap_atomid[k]].resind;
1457                 if (0 == k)
1458                 {
1459                     cmap_chainnum = resinfo[this_residue_index].chainnum;
1460                 }
1461                 else
1462                 {
1463                     /* Does the residue for this atom have the same
1464                      * chain number as the residues for previous
1465                      * atoms? */
1466                     bAddCMAP = bAddCMAP && cmap_chainnum == resinfo[this_residue_index].chainnum;
1467                 }
1468                 /* Here we used to check that the residuetype was protein and
1469                  * disable bAddCMAP if that was not the case. However, some
1470                  * special residues (say, alanine dipeptides) might not adhere
1471                  * to standard naming, and if we start calling them normal
1472                  * protein residues the user will be bugged to select termini.
1473                  *
1474                  * Instead, I believe that the right course of action is to
1475                  * keep the CMAP interaction if it is present in the RTP file
1476                  * and we correctly identified all atoms (which is the case
1477                  * if we got here).
1478                  */
1479             }
1480
1481             if (bAddCMAP)
1482             {
1483                 add_cmap_param(psb,
1484                                cmap_atomid[0],
1485                                cmap_atomid[1],
1486                                cmap_atomid[2],
1487                                cmap_atomid[3],
1488                                cmap_atomid[4],
1489                                b.s.c_str());
1490             }
1491         }
1492
1493         if (residx < nres - 1)
1494         {
1495             while (atoms->atom[i].resind < residx + 1)
1496             {
1497                 i++;
1498             }
1499         }
1500     }
1501     /* Start the next residue */
1502 }
1503
1504 static void scrub_charge_groups(int* cgnr, int natoms)
1505 {
1506     int i;
1507
1508     for (i = 0; i < natoms; i++)
1509     {
1510         cgnr[i] = i + 1;
1511     }
1512 }
1513
1514
1515 void pdb2top(FILE*                                  top_file,
1516              const char*                            posre_fn,
1517              const char*                            molname,
1518              t_atoms*                               atoms,
1519              std::vector<gmx::RVec>*                x,
1520              PreprocessingAtomTypes*                atype,
1521              t_symtab*                              tab,
1522              gmx::ArrayRef<const PreprocessResidue> rtpFFDB,
1523              gmx::ArrayRef<PreprocessResidue>       usedPpResidues,
1524              gmx::ArrayRef<MoleculePatchDatabase>   globalPatches,
1525              bool                                   bAllowMissing,
1526              bool                                   bVsites,
1527              bool                                   bVsiteAromatics,
1528              const char*                            ffdir,
1529              real                                   mHmult,
1530              gmx::ArrayRef<const DisulfideBond>     ssbonds,
1531              real                                   long_bond_dist,
1532              real                                   short_bond_dist,
1533              bool                                   bDeuterate,
1534              bool                                   bChargeGroups,
1535              bool                                   bCmap,
1536              bool                                   bRenumRes,
1537              bool                                   bRTPresname,
1538              gmx::ArrayRef<const int>               cyclicBondsIndex,
1539              const gmx::MDLogger&                   logger)
1540 {
1541     std::array<InteractionsOfType, F_NRE> plist;
1542     t_excls*                              excls;
1543     int*                                  cgnr;
1544     int*                                  vsite_type;
1545     int                                   i, nmissat;
1546     int                                   bts[ebtsNR];
1547
1548     ResidueType rt;
1549
1550     /* Make bonds */
1551     at2bonds(&(plist[F_BONDS]), globalPatches, atoms, *x, long_bond_dist, short_bond_dist, cyclicBondsIndex, logger);
1552
1553     /* specbonds: disulphide bonds & heme-his */
1554     do_ssbonds(&(plist[F_BONDS]), atoms, ssbonds, bAllowMissing);
1555
1556     nmissat = name2type(atoms, &cgnr, usedPpResidues, &rt, logger);
1557     if (nmissat)
1558     {
1559         if (bAllowMissing)
1560         {
1561             GMX_LOG(logger.warning)
1562                     .asParagraph()
1563                     .appendTextFormatted("There were %d missing atoms in molecule %s", nmissat, molname);
1564         }
1565         else
1566         {
1567             gmx_fatal(FARGS,
1568                       "There were %d missing atoms in molecule %s, if you want to use this "
1569                       "incomplete topology anyhow, use the option -missing",
1570                       nmissat,
1571                       molname);
1572         }
1573     }
1574
1575     /* Cleanup bonds (sort and rm doubles) */
1576     clean_bonds(&(plist[F_BONDS]), logger);
1577
1578     snew(vsite_type, atoms->nr);
1579     for (i = 0; i < atoms->nr; i++)
1580     {
1581         vsite_type[i] = NOTSET;
1582     }
1583     if (bVsites)
1584     {
1585         if (bVsiteAromatics)
1586         {
1587             GMX_LOG(logger.info)
1588                     .asParagraph()
1589                     .appendTextFormatted(
1590                             "The conversion of aromatic rings into virtual sites is deprecated "
1591                             "and may be removed in a future version of GROMACS");
1592         }
1593         /* determine which atoms will be vsites and add dummy masses
1594            also renumber atom numbers in plist[0..F_NRE]! */
1595         do_vsites(rtpFFDB, atype, atoms, tab, x, plist, &vsite_type, &cgnr, mHmult, bVsiteAromatics, ffdir);
1596     }
1597
1598     /* Make Angles and Dihedrals */
1599     GMX_LOG(logger.info)
1600             .asParagraph()
1601             .appendTextFormatted("Generating angles, dihedrals and pairs...");
1602     snew(excls, atoms->nr);
1603     gen_pad(atoms, usedPpResidues, plist, excls, globalPatches, bAllowMissing, cyclicBondsIndex);
1604
1605     /* Make CMAP */
1606     if (bCmap)
1607     {
1608         gen_cmap(&(plist[F_CMAP]), usedPpResidues, atoms, cyclicBondsIndex, logger);
1609         if (plist[F_CMAP].size() > 0)
1610         {
1611             GMX_LOG(logger.info)
1612                     .asParagraph()
1613                     .appendTextFormatted("There are %4zu cmap torsion pairs", plist[F_CMAP].size());
1614         }
1615     }
1616
1617     /* set mass of all remaining hydrogen atoms */
1618     if (mHmult != 1.0)
1619     {
1620         do_h_mass(&(plist[F_BONDS]), vsite_type, atoms, mHmult, bDeuterate);
1621     }
1622     sfree(vsite_type);
1623
1624     /* Cleanup bonds (sort and rm doubles) */
1625     /* clean_bonds(&(plist[F_BONDS]));*/
1626
1627     GMX_LOG(logger.info)
1628             .asParagraph()
1629             .appendTextFormatted(
1630                     "There are %4zu dihedrals, %4zu impropers, %4zu angles\n"
1631                     "          %4zu pairs,     %4zu bonds and  %4zu virtual sites",
1632                     plist[F_PDIHS].size(),
1633                     plist[F_IDIHS].size(),
1634                     plist[F_ANGLES].size(),
1635                     plist[F_LJ14].size(),
1636                     plist[F_BONDS].size(),
1637                     plist[F_VSITE2].size() + plist[F_VSITE3].size() + plist[F_VSITE3FD].size()
1638                             + plist[F_VSITE3FAD].size() + plist[F_VSITE3OUT].size()
1639                             + plist[F_VSITE4FD].size() + plist[F_VSITE4FDN].size());
1640
1641     print_sums(atoms, FALSE, logger);
1642
1643     if (!bChargeGroups)
1644     {
1645         scrub_charge_groups(cgnr, atoms->nr);
1646     }
1647
1648     if (bRenumRes)
1649     {
1650         for (i = 0; i < atoms->nres; i++)
1651         {
1652             atoms->resinfo[i].nr = i + 1;
1653             atoms->resinfo[i].ic = ' ';
1654         }
1655     }
1656
1657     if (top_file)
1658     {
1659         GMX_LOG(logger.info).asParagraph().appendTextFormatted("Writing topology");
1660         /* We can copy the bonded types from the first restp,
1661          * since the types have to be identical for all residues in one molecule.
1662          */
1663         for (i = 0; i < ebtsNR; i++)
1664         {
1665             bts[i] = usedPpResidues[0].rb[i].type;
1666         }
1667         write_top(top_file,
1668                   posre_fn,
1669                   molname,
1670                   atoms,
1671                   bRTPresname,
1672                   bts,
1673                   plist,
1674                   excls,
1675                   atype,
1676                   cgnr,
1677                   usedPpResidues[0].nrexcl);
1678     }
1679
1680
1681     /* we should clean up hb and restp here, but that is a *L*O*T* of work! */
1682     sfree(cgnr);
1683     for (i = 0; i < atoms->nr; i++)
1684     {
1685         sfree(excls[i].e);
1686     }
1687     sfree(excls);
1688 }