src/kernel/nm2type.c

   1 /*
   2  *
   3  *                This source code is part of
   4  *
   5  *                 G   R   O   M   A   C   S
   6  *
   7  *          GROningen MAchine for Chemical Simulations
   8  *
   9  *                        VERSION 3.3.3
  10  * Written by David van der Spoel, Erik Lindahl, Berk Hess, and others.
  11  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
  12  * Copyright (c) 2001-2008, The GROMACS development team,
  13  * check out http://www.gromacs.org for more information.
  14
  15  * This program is free software; you can redistribute it and/or
  16  * modify it under the terms of the GNU General Public License
  17  * as published by the Free Software Foundation; either version 2
  18  * of the License, or (at your option) any later version.
  19  *
  20  * If you want to redistribute modifications, please consider that
  21  * scientific software is very special. Version control is crucial -
  22  * bugs must be traceable. We will be happy to consider code for
  23  * inclusion in the official distribution, but derived work must not
  24  * be called official GROMACS. Details are found in the README & COPYING
  25  * files - if they are missing, get the official version at www.gromacs.org.
  26  *
  27  * To help us fund GROMACS development, we humbly ask that you cite
  28  * the papers on the package - you can find them in the top README file.
  29  *
  30  * For more info, check our website at http://www.gromacs.org
  31  *
  32  * And Hey:
  33  * Groningen Machine for Chemical Simulation
  34  */
  35 /* This file is completely threadsafe - keep it that way! */
  36 #ifdef HAVE_CONFIG_H
  37 #include <config.h>
  38 #endif
  39
  40 #include "assert.h"
  41 #include "maths.h"
  42 #include "macros.h"
  43 #include "copyrite.h"
  44 #include "bondf.h"
  45 #include "string2.h"
  46 #include "smalloc.h"
  47 #include "sysstuff.h"
  48 #include "confio.h"
  49 #include "physics.h"
  50 #include "statutil.h"
  51 #include "vec.h"
  52 #include "random.h"
  53 #include "3dview.h"
  54 #include "txtdump.h"
  55 #include "readinp.h"
  56 #include "names.h"
  57 #include "toppush.h"
  58 #include "pdb2top.h"
  59 #include "gpp_nextnb.h"
  60 #include "gpp_atomtype.h"
  61 #include "x2top.h"
  62 #include "fflibutil.h"
  63
  64 static void rd_nm2type_file(const char *fn,int *nnm,t_nm2type **nmp)
  65 {
  66   FILE      *fp;
  67   bool      bCont;
  68   char      libfilename[128];
  69   char      format[128],f1[128];
  70   char      buf[1024],elem[16],type[16],nbbuf[16],**newbuf;
  71   int       i,nb,nnnm,line=1;
  72   double    qq,mm,*blen;
  73   t_nm2type *nm2t=NULL;
  74
  75   fp = fflib_open(fn);
  76
  77   nnnm = *nnm;
  78   nm2t = *nmp;
  79   do {
  80     /* Read a line from the file */
  81     bCont = (fgets2(buf,1023,fp) != NULL);
  82
  83     if (bCont) {
  84       /* Remove comment */
  85       strip_comment(buf);
  86       if (sscanf(buf,"%s%s%lf%lf%d",elem,type,&qq,&mm,&nb) == 5) {
  87         /* If we can read the first four, there probably is more */
  88         srenew(nm2t,nnnm+1);
  89         snew(nm2t[nnnm].blen,nb);
  90         if (nb > 0) {
  91           snew(newbuf,nb);
  92           strcpy(format,"%*s%*s%*s%*s%*s");
  93           for(i=0; (i<nb); i++) {
  94             /* Complicated format statement */
  95             strcpy(f1,format);
  96             strcat(f1,"%s%lf");
  97             if (sscanf(buf,f1,nbbuf,&(nm2t[nnnm].blen[i])) != 2)
  98               gmx_fatal(FARGS,"Error on line %d of %s",line,libfilename);
  99             newbuf[i] = strdup(nbbuf);
 100             strcat(format,"%*s%*s");
 101           }
 102         }
 103         else
 104           newbuf = NULL;
 105         nm2t[nnnm].elem   = strdup(elem);
 106         nm2t[nnnm].type   = strdup(type);
 107         nm2t[nnnm].q      = qq;
 108         nm2t[nnnm].m      = mm;
 109         nm2t[nnnm].nbonds = nb;
 110         nm2t[nnnm].bond   = newbuf;
 111         nnnm++;
 112       }
 113       line++;
 114     }
 115   } while(bCont);
 116   ffclose(fp);
 117
 118   *nnm = nnnm;
 119   *nmp = nm2t;
 120 }
 121
 122 t_nm2type *rd_nm2type(const char *ffdir,int *nnm)
 123 {
 124   int  nff,f;
 125   char **ff;
 126   t_nm2type *nm;
 127
 128   nff = fflib_search_file_end(ffdir,".n2t",FALSE,&ff);
 129   *nnm = 0;
 130   nm   = NULL;
 131   for(f=0; f<nff; f++) {
 132     rd_nm2type_file(ff[f],nnm,&nm);
 133     sfree(ff[f]);
 134   }
 135   sfree(ff);
 136
 137   return nm;
 138 }
 139
 140 void dump_nm2type(FILE *fp,int nnm,t_nm2type nm2t[])
 141 {
 142   int i,j;
 143
 144   fprintf(fp,"; nm2type database\n");
 145   for(i=0; (i<nnm); i++) {
 146     fprintf(fp,"%-8s %-8s %8.4f %8.4f %-4d",
 147             nm2t[i].elem,nm2t[i].type,
 148             nm2t[i].q,nm2t[i].m,nm2t[i].nbonds);
 149     for(j=0; (j<nm2t[i].nbonds); j++)
 150       fprintf(fp," %-5s %6.4f",nm2t[i].bond[j],nm2t[i].blen[j]);
 151     fprintf(fp,"\n");
 152   }
 153 }
 154
 155 enum { ematchNone, ematchWild, ematchElem, ematchExact, ematchNR };
 156
 157 static int match_str(const char *atom,const char *template_string)
 158 {
 159   if (!atom || !template_string)
 160     return ematchNone;
 161   else if (strcasecmp(atom,template_string) == 0)
 162     return ematchExact;
 163   else if (atom[0] == template_string[0])
 164     return ematchElem;
 165   else if (strcmp(template_string,"*") == 0)
 166     return ematchWild;
 167   else
 168     return ematchNone;
 169 }
 170
 171 int nm2type(int nnm,t_nm2type nm2t[],t_symtab *tab,t_atoms *atoms,
 172             gpp_atomtype_t atype,int *nbonds,t_params *bonds)
 173 {
 174   int cur = 0;
 175 #define prev (1-cur)
 176   int i,j,k,m,n,nresolved,nb,maxbond,ai,aj,best,im,nqual[2][ematchNR];
 177   int *bbb,*n_mask,*m_mask,**match,**quality;
 178   char *aname_i,*aname_m,*aname_n,*type;
 179   double qq,mm;
 180   t_param *param;
 181
 182   snew(param,1);
 183   maxbond = 0;
 184   for(i=0; (i<atoms->nr); i++)
 185     maxbond = max(maxbond,nbonds[i]);
 186   if (debug)
 187     fprintf(debug,"Max number of bonds per atom is %d\n",maxbond);
 188   snew(bbb,maxbond);
 189   snew(n_mask,maxbond);
 190   snew(m_mask,maxbond);
 191   snew(match,maxbond);
 192   for(i=0; (i<maxbond); i++)
 193     snew(match[i],maxbond);
 194
 195   nresolved = 0;
 196   for(i=0; (i<atoms->nr); i++) {
 197     aname_i = *atoms->atomname[i];
 198     nb = 0;
 199     for(j=0; (j<bonds->nr); j++) {
 200       ai = bonds->param[j].AI;
 201       aj = bonds->param[j].AJ;
 202       if (ai == i)
 203         bbb[nb++] = aj;
 204       else if (aj == i)
 205         bbb[nb++] = ai;
 206     }
 207     if (nb != nbonds[i])
 208       gmx_fatal(FARGS,"Counting number of bonds nb = %d, nbonds[%d] = %d",
 209                 nb,i,nbonds[i]);
 210     if(debug) {
 211       fprintf(debug,"%4s has bonds to",aname_i);
 212       for(j=0; (j<nb); j++)
 213         fprintf(debug," %4s",*atoms->atomname[bbb[j]]);
 214       fprintf(debug,"\n");
 215     }
 216     best = -1;
 217     for(k=0; (k<ematchNR); k++)
 218       nqual[prev][k] = 0;
 219
 220     /* First check for names */
 221     for(k=0; (k<nnm); k++) {
 222       if (nm2t[k].nbonds == nb) {
 223         im = match_str(*atoms->atomname[i],nm2t[k].elem);
 224         if (im > ematchWild) {
 225           for(j=0; (j<ematchNR); j++)
 226             nqual[cur][j] = 0;
 227
 228           /* Fill a matrix with matching quality */
 229           for(m=0; (m<nb); m++) {
 230             aname_m = *atoms->atomname[bbb[m]];
 231             for(n=0; (n<nb); n++) {
 232               aname_n = nm2t[k].bond[n];
 233               match[m][n] = match_str(aname_m,aname_n);
 234             }
 235           }
 236           /* Now pick the best matches */
 237           for(m=0; (m<nb); m++) {
 238             n_mask[m] = 0;
 239             m_mask[m] = 0;
 240           }
 241           for(j=ematchNR-1; (j>0); j--) {
 242             for(m=0; (m<nb); m++) {
 243               for(n=0; (n<nb); n++) {
 244                 if ((n_mask[n] == 0) &&
 245                     (m_mask[m] == 0) &&
 246                     (match[m][n] == j)) {
 247                   n_mask[n] = 1;
 248                   m_mask[m] = 1;
 249                   nqual[cur][j]++;
 250                 }
 251               }
 252             }
 253           }
 254           if ((nqual[cur][ematchExact]+
 255                nqual[cur][ematchElem]+
 256                nqual[cur][ematchWild]) == nb) {
 257             if ((nqual[cur][ematchExact] > nqual[prev][ematchExact]) ||
 258
 259                 ((nqual[cur][ematchExact] == nqual[prev][ematchExact]) &&
 260                  (nqual[cur][ematchElem] > nqual[prev][ematchElem])) ||
 261
 262                 ((nqual[cur][ematchExact] == nqual[prev][ematchExact]) &&
 263                  (nqual[cur][ematchElem] == nqual[prev][ematchElem]) &&
 264                  (nqual[cur][ematchWild] > nqual[prev][ematchWild]))) {
 265               best = k;
 266               cur  = prev;
 267             }
 268           }
 269         }
 270       }
 271     }
 272     if (best != -1) {
 273       int  atomnr=0;
 274       real alpha=0;
 275
 276       qq   = nm2t[best].q;
 277       mm   = nm2t[best].m;
 278       type = nm2t[best].type;
 279
 280       if ((k = get_atomtype_type(type,atype)) == NOTSET) {
 281         atoms->atom[i].qB = alpha;
 282         atoms->atom[i].m = atoms->atom[i].mB = mm;
 283         k = add_atomtype(atype,tab,&(atoms->atom[i]),type,param,
 284                          atoms->atom[i].type,0,0,0,atomnr,0,0);
 285       }
 286       atoms->atom[i].type  = k;
 287       atoms->atom[i].typeB = k;
 288       atoms->atom[i].q  = qq;
 289       atoms->atom[i].qB = qq;
 290       atoms->atom[i].m  = mm;
 291       atoms->atom[i].mB = mm;
 292       nresolved++;
 293     }
 294     else {
 295       fprintf(stderr,"Can not find forcefield for atom %s-%d with %d bonds\n",
 296               *atoms->atomname[i],i+1,nb);
 297     }
 298   }
 299   sfree(bbb);
 300   sfree(n_mask);
 301   sfree(m_mask);
 302   sfree(param);
 303
 304   return nresolved;
 305 }
 306