pcacommand.cpp

   1
   2 /*
   3  *  pcacommand.cpp
   4  *  Mothur
   5  *
   6  *  Created by westcott on 1/4/10.
   7  *  Copyright 2010 Schloss Lab. All rights reserved.
   8  *
   9  */
  10
  11 #include "pcacommand.h"
  12
  13 //**********************************************************************************************************************
  14
  15 PCACommand::PCACommand(string option){
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"phylip","lt"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string, string> parameters = parser. getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //required parameters
  38                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  39                         if (phylipfile == "not open") { abort = true; }
  40                         else if (phylipfile == "not found") { phylipfile = ""; abort = true; }
  41                         else {  filename = phylipfile;  }
  42
  43                         //columnfile = validParameter.validFile(parameters, "column", true);
  44                         //if (columnfile == "not open") { abort = true; }
  45                         //else if (columnfile == "not found") { columnfile = ""; }
  46                         //else {  format = "column";    }
  47
  48                         //namefile = validParameter.validFile(parameters, "name", true);
  49                         //if (namefile == "not open") { abort = true; }
  50                         //else if (namefile == "not found") { namefile = ""; }
  51
  52
  53                         //error checking on files
  54                         if (phylipfile == "")   { mothurOut("You must provide a distance file before running the pca command."); mothurOutEndLine(); abort = true; }
  55                         //if ((phylipfile == "") && (columnfile == "")) { mothurOut("You must provide a distance file before running the pca command."); mothurOutEndLine(); abort = true; }
  56                         //else if ((phylipfile != "") && (columnfile != "")) { mothurOut("You may not use both the column and the phylip parameters."); mothurOutEndLine(); abort = true; }
  57
  58                         //if (columnfile != "") {
  59                         //      if (namefile == "") {  mothurOut("You need to provide a namefile if you are going to use the column format."); mothurOutEndLine(); abort = true; }
  60                         //}
  61
  62                         string temp = validParameter.validFile(parameters, "lt", false);                                if (temp == "not found") { temp = "false"; }
  63                         bool lt = isTrue(temp);
  64
  65                         if (lt)         {  matrix = 2;  }
  66                         else            {  matrix = 1;  }
  67
  68
  69                 }
  70
  71         }
  72         catch(exception& e) {
  73                 errorOut(e, "PCACommand", "PCACommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 void PCACommand::help(){
  79         try {
  80
  81                 mothurOut("The pca command..."); mothurOutEndLine();
  82         }
  83         catch(exception& e) {
  84                 errorOut(e, "PCACommand", "help");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 PCACommand::~PCACommand(){}
  90 //**********************************************************************************************************************
  91 int PCACommand::execute(){
  92         try {
  93
  94                 if (abort == true) { return 0; }
  95
  96                 cout.setf(ios::fixed, ios::floatfield);
  97                 cout.setf(ios::showpoint);
  98                 cerr.setf(ios::fixed, ios::floatfield);
  99                 cerr.setf(ios::showpoint);
 100
 101                 vector<string> names;
 102                 vector<vector<double> > D;
 103
 104                 fbase = filename;
 105                 if(fbase.find_last_of(".")!=string::npos){
 106                         fbase.erase(fbase.find_last_of(".")+1);
 107                 }
 108                 else{
 109                         fbase += ".";
 110                 }
 111                 read(filename, matrix, names, D);
 112
 113                 double offset = 0.0000;
 114                 vector<double> d;
 115                 vector<double> e;
 116                 vector<vector<double> > G = D;
 117                 vector<vector<double> > copy_G;
 118                 int rank = D.size();
 119
 120                 cout << "\nProcessing...\n";
 121
 122                 for(int count=0;count<2;count++){
 123                         recenter(offset, D, G);
 124                         tred2(G, d, e);
 125                         qtli(d, e, G);
 126                         offset = d[d.size()-1];
 127                         if(offset > 0.0) break;
 128                 }
 129
 130
 131                 output(fbase, names, G, d);
 132
 133                 return 0;
 134         }
 135         catch(exception& e) {
 136                 errorOut(e, "PCACommand", "execute");
 137                 exit(1);
 138         }
 139 }
 140 /*********************************************************************************************************************************/
 141
 142 inline double SIGN(const double a, const double b)
 143 {
 144     return b>=0 ? (a>=0 ? a:-a) : (a>=0 ? -a:a);
 145 }
 146
 147 /*********************************************************************************************************************************/
 148
 149 void PCACommand::get_comment(istream& f, char begin, char end){
 150         try {
 151                 char d=f.get();
 152                 while(d != end){        d = f.get();    }
 153                 d = f.peek();
 154         }
 155         catch(exception& e) {
 156                 errorOut(e, "PCACommand", "get_comment");
 157                 exit(1);
 158         }
 159 }
 160
 161 /*********************************************************************************************************************************/
 162
 163 void PCACommand::read_mega(istream& f, int square_m, vector<string>& name_list, vector<vector<double> >& d){
 164         try {
 165                 get_comment(f, '#', '\n');
 166
 167                 char test = f.peek();
 168
 169                 while(test == '!'){                                                             //get header comments
 170                         get_comment(f, '!', ';');
 171                         while(isspace(test=f.get()))            {;}
 172                         f.putback(test);
 173                         test = f.peek();
 174                 }
 175                 while(test != '\n'){                                                    //get sequence names
 176                         get_comment(f, '[', ']');
 177                         char d = f.get();
 178                         d = f.get();
 179                         if(d == '#'){
 180                                 string name;
 181                                 f >> name;
 182                                 name_list.push_back(name);
 183                                 while(isspace(test=f.get()))            {;}
 184                                 f.putback(test);
 185                         }
 186                         else{
 187                                 break;
 188                         }
 189                 }
 190                 int rank = name_list.size();
 191                 d.resize(rank);
 192                 for(int i=0;i<rank;i++){                d[i].resize(rank);              }
 193
 194                 d[0][0] = 0.0000;
 195                 get_comment(f, '[', ']');
 196                 for(int i=1;i<rank;i++){
 197                         get_comment(f, '[', ']');
 198                         d[i][i]=0.0000;
 199                         for(int j=0;j<i;j++){
 200                                 f >> d[i][j];
 201                                 if (d[i][j] == -0.0000)
 202                                         d[i][j] = 0.0000;
 203                                 d[j][i]=d[i][j];
 204                         }
 205                 }
 206         }
 207         catch(exception& e) {
 208                 errorOut(e, "PCACommand", "read_mega");
 209                 exit(1);
 210         }
 211 }
 212
 213 /*********************************************************************************************************************************/
 214
 215 void PCACommand::read_phylip(istream& f, int square_m, vector<string>& name_list, vector<vector<double> >& d){
 216         try {
 217                 //     int count1=0;
 218                 //     int count2=0;
 219
 220                 int rank;
 221                 f >> rank;
 222
 223                 name_list.resize(rank);
 224                 d.resize(rank);
 225                 if(square_m == 1){
 226                         for(int i=0;i<rank;i++)
 227                                 d[i].resize(rank);
 228                         for(int i=0;i<rank;i++) {
 229                                 f >> name_list[i];
 230                                 //                      cout << i << "\t" << name_list[i] << endl;
 231                                 for(int j=0;j<rank;j++) {
 232                                         f >> d[i][j];
 233                                         if (d[i][j] == -0.0000)
 234                                                 d[i][j] = 0.0000;
 235                                 }
 236                         }
 237                 }
 238                 else if(square_m == 2){
 239                         for(int i=0;i<rank;i++){
 240                                 d[i].resize(rank);
 241                         }
 242                         d[0][0] = 0.0000;
 243                         f >> name_list[0];
 244                         for(int i=1;i<rank;i++){
 245                                 f >> name_list[i];
 246                                 d[i][i]=0.0000;
 247                                 for(int j=0;j<i;j++){
 248                                         f >> d[i][j];
 249                                         if (d[i][j] == -0.0000)
 250                                                 d[i][j] = 0.0000;
 251                                         d[j][i]=d[i][j];
 252                                 }
 253                         }
 254                 }
 255         }
 256         catch(exception& e) {
 257                 errorOut(e, "PCACommand", "read_phylip");
 258                 exit(1);
 259         }
 260
 261 }
 262
 263 /*********************************************************************************************************************************/
 264
 265 void PCACommand::read(string fname, int m, vector<string>& names, vector<vector<double> >& D){
 266         try {
 267                 ifstream f(fname.c_str());
 268                 if(!f) {
 269                         cerr << "Error: Could not open " << fname << endl;
 270                         exit(1);
 271                 }
 272                 char test = f.peek();
 273
 274                 if(test == '#'){
 275                         read_mega(f, m, names, D);
 276                 }
 277                 else{
 278                         read_phylip(f, m, names, D);
 279                 }
 280
 281                 int rank = D.size();
 282         }
 283         catch(exception& e) {
 284                 errorOut(e, "PCACommand", "read");
 285                 exit(1);
 286         }
 287 }
 288
 289 /*********************************************************************************************************************************/
 290 double PCACommand::pythag(double a, double b){
 291     return(pow(a*a+b*b,0.5));
 292 }
 293 /*********************************************************************************************************************************/
 294
 295 void PCACommand::matrix_mult(vector<vector<double> > first, vector<vector<double> > second, vector<vector<double> >& product){
 296         try {
 297                 int first_rows = first.size();
 298                 int first_cols = first[0].size();
 299                 int second_cols = second[0].size();
 300
 301                 product.resize(first_rows);
 302                 for(int i=0;i<first_rows;i++){
 303                         product[i].resize(second_cols);
 304                 }
 305
 306                 for(int i=0;i<first_rows;i++){
 307                         for(int j=0;j<second_cols;j++){
 308                                 product[i][j] = 0.0;
 309                                 for(int k=0;k<first_cols;k++){
 310                                         product[i][j] += first[i][k] * second[k][j];
 311                                 }
 312                         }
 313                 }
 314         }
 315         catch(exception& e) {
 316                 errorOut(e, "PCACommand", "matrix_mult");
 317                 exit(1);
 318         }
 319
 320 }
 321
 322 /*********************************************************************************************************************************/
 323
 324 void PCACommand::recenter(double offset, vector<vector<double> > D, vector<vector<double> >& G){
 325         try {
 326                 int rank = D.size();
 327
 328                 vector<vector<double> > A(rank);
 329                 vector<vector<double> > C(rank);
 330                 for(int i=0;i<rank;i++){
 331                         A[i].resize(rank);
 332                         C[i].resize(rank);
 333                 }
 334
 335                 double scale = -1.0000 / (double) rank;
 336
 337                 for(int i=0;i<rank;i++){
 338                         A[i][i] = 0.0000;
 339                         C[i][i] = 1.0000 + scale;
 340                         for(int j=i+1;j<rank;j++){
 341                                 A[i][j] = A[j][i] = -0.5 * D[i][j] * D[i][j] + offset;
 342                                 C[i][j] = C[j][i] = scale;
 343                         }
 344                 }
 345
 346                 matrix_mult(C,A,A);
 347                 matrix_mult(A,C,G);
 348         }
 349         catch(exception& e) {
 350                 errorOut(e, "PCACommand", "recenter");
 351                 exit(1);
 352         }
 353
 354 }
 355
 356 /*********************************************************************************************************************************/
 357
 358 //  This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
 359
 360 void PCACommand::tred2(vector<vector<double> >& a, vector<double>& d, vector<double>& e){
 361         try {
 362                 double scale, hh, h, g, f;
 363
 364                 int n = a.size();
 365
 366                 d.resize(n);
 367                 e.resize(n);
 368
 369                 for(int i=n-1;i>0;i--){
 370                         int l=i-1;
 371                         h = scale = 0.0000;
 372                         if(l>0){
 373                                 for(int k=0;k<l+1;k++){
 374                                         scale += fabs(a[i][k]);
 375                                 }
 376                                 if(scale == 0.0){
 377                                         e[i] = a[i][l];
 378                                 }
 379                                 else{
 380                                         for(int k=0;k<l+1;k++){
 381                                                 a[i][k] /= scale;
 382                                                 h += a[i][k] * a[i][k];
 383                                         }
 384                                         f = a[i][l];
 385                                         g = (f >= 0.0 ? -sqrt(h) : sqrt(h));
 386                                         e[i] = scale * g;
 387                                         h -= f * g;
 388                                         a[i][l] = f - g;
 389                                         f = 0.0;
 390                                         for(int j=0;j<l+1;j++){
 391                                                 a[j][i] = a[i][j] / h;
 392                                                 g = 0.0;
 393                                                 for(int k=0;k<j+1;k++){
 394                                                         g += a[j][k] * a[i][k];
 395                                                 }
 396                                                 for(int k=j+1;k<l+1;k++){
 397                                                         g += a[k][j] * a[i][k];
 398                                                 }
 399                                                 e[j] = g / h;
 400                                                 f += e[j] * a[i][j];
 401                                         }
 402                                         hh = f / (h + h);
 403                                         for(int j=0;j<l+1;j++){
 404                                                 f = a[i][j];
 405                                                 e[j] = g = e[j] - hh * f;
 406                                                 for(int k=0;k<j+1;k++){
 407                                                         a[j][k] -= (f * e[k] + g * a[i][k]);
 408                                                 }
 409                                         }
 410                                 }
 411                         }
 412                         else{
 413                                 e[i] = a[i][l];
 414                         }
 415
 416                         d[i] = h;
 417                 }
 418
 419                 d[0] = 0.0000;
 420                 e[0] = 0.0000;
 421
 422                 for(int i=0;i<n;i++){
 423                         int l = i;
 424                         if(d[i] != 0.0){
 425                                 for(int j=0;j<l;j++){
 426                                         g = 0.0000;
 427                                         for(int k=0;k<l;k++){
 428                                                 g += a[i][k] * a[k][j];
 429                                         }
 430                                         for(int k=0;k<l;k++){
 431                                                 a[k][j] -= g * a[k][i];
 432                                         }
 433                                 }
 434                         }
 435                         d[i] = a[i][i];
 436                         a[i][i] = 1.0000;
 437                         for(int j=0;j<l;j++){
 438                                 a[j][i] = a[i][j] = 0.0;
 439                         }
 440                 }
 441         }
 442         catch(exception& e) {
 443                 errorOut(e, "PCACommand", "tred2");
 444                 exit(1);
 445         }
 446
 447 }
 448
 449 /*********************************************************************************************************************************/
 450
 451 //  This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
 452
 453 void PCACommand::qtli(vector<double>& d, vector<double>& e, vector<vector<double> >& z) {
 454         try {
 455                 int m, i, iter;
 456                 double s, r, p, g, f, dd, c, b;
 457
 458                 int n = d.size();
 459                 for(int i=1;i<=n;i++){
 460                         e[i-1] = e[i];
 461                 }
 462                 e[n-1] = 0.0000;
 463
 464                 for(int l=0;l<n;l++){
 465                         iter = 0;
 466                         do {
 467                                 for(m=l;m<n-1;m++){
 468                                         dd = fabs(d[m]) + fabs(d[m+1]);
 469                                         if(fabs(e[m])+dd == dd) break;
 470                                 }
 471                                 if(m != l){
 472                                         if(iter++ == 30) cerr << "Too many iterations in tqli\n";
 473                                         g = (d[l+1]-d[l]) / (2.0 * e[l]);
 474                                         r = pythag(g, 1.0);
 475                                         g = d[m] - d[l] + e[l] / (g + SIGN(r,g));
 476                                         s = c = 1.0;
 477                                         p = 0.0000;
 478                                         for(i=m-1;i>=l;i--){
 479                                                 f = s * e[i];
 480                                                 b = c * e[i];
 481                                                 e[i+1] = (r=pythag(f,g));
 482                                                 if(r==0.0){
 483                                                         d[i+1] -= p;
 484                                                         e[m] = 0.0000;
 485                                                         break;
 486                                                 }
 487                                                 s = f / r;
 488                                                 c = g / r;
 489                                                 g = d[i+1] - p;
 490                                                 r = (d[i] - g) * s + 2.0 * c * b;
 491                                                 d[i+1] = g + ( p = s * r);
 492                                                 g = c * r - b;
 493                                                 for(int k=0;k<n;k++){
 494                                                         f = z[k][i+1];
 495                                                         z[k][i+1] = s * z[k][i] + c * f;
 496                                                         z[k][i] = c * z[k][i] - s * f;
 497                                                 }
 498                                         }
 499                                         if(r == 0.00 && i >= l) continue;
 500                                         d[l] -= p;
 501                                         e[l] = g;
 502                                         e[m] = 0.0;
 503                                 }
 504                         } while (m != l);
 505                 }
 506
 507                 int k;
 508                 for(int i=0;i<n;i++){
 509                         p=d[k=i];
 510                         for(int j=i;j<n;j++){
 511                                 if(d[j] >= p){
 512                                         p=d[k=j];
 513                                 }
 514                         }
 515                         if(k!=i){
 516                                 d[k]=d[i];
 517                                 d[i]=p;
 518                                 for(int j=0;j<n;j++){
 519                                         p=z[j][i];
 520                                         z[j][i] = z[j][k];
 521                                         z[j][k] = p;
 522                                 }
 523                         }
 524                 }
 525         }
 526         catch(exception& e) {
 527                 errorOut(e, "PCACommand", "qtli");
 528                 exit(1);
 529         }
 530 }
 531
 532 /*********************************************************************************************************************************/
 533
 534 void PCACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> > G, vector<double> d) {
 535         try {
 536                 int rank = name_list.size();
 537                 double dsum = 0.0000;
 538                 for(int i=0;i<rank;i++){
 539                         dsum += d[i];
 540                         for(int j=0;j<rank;j++){
 541                                 if(d[j] >= 0)   {       G[i][j] *= pow(d[j],0.5);       }
 542                                 else                    {       G[i][j] = 0.00000;                      }
 543                         }
 544                 }
 545
 546                 ofstream pcaData((fnameRoot+"pca").c_str(), ios::trunc);
 547                 pcaData.setf(ios::fixed, ios::floatfield);
 548                 pcaData.setf(ios::showpoint);
 549
 550                 ofstream pcaLoadings((fnameRoot+"pca.loadings").c_str(), ios::trunc);
 551                 pcaLoadings.setf(ios::fixed, ios::floatfield);
 552                 pcaLoadings.setf(ios::showpoint);
 553
 554                 pcaLoadings << "axis\tloading\n";
 555                 for(int i=0;i<rank;i++){
 556                         pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
 557                 }
 558
 559                 pcaData << "SeqName";
 560                 for(int i=0;i<rank;i++){
 561                         pcaData << '\t' << "axis" << i+1;
 562                 }
 563                 pcaData << endl;
 564
 565                 for(int i=0;i<rank;i++){
 566                         pcaData << name_list[i] << '\t';
 567                         for(int j=0;j<rank;j++){
 568                                 pcaData << G[i][j] << '\t';
 569                         }
 570                         pcaData << endl;
 571                 }
 572         }
 573         catch(exception& e) {
 574                 errorOut(e, "PCACommand", "output");
 575                 exit(1);
 576         }
 577 }
 578
 579 /*********************************************************************************************************************************/
 580
 581 void PCACommand::print_matrix(vector<vector<double> > A) {
 582         try {
 583                 int rank = A.size();
 584                 for(int i=0;i<rank;i++){
 585                         for(int j=0;j<rank;j++){
 586                                 cout << A[i][j] << "  ";
 587                         }
 588                         cout << endl;
 589                 }
 590         }
 591         catch(exception& e) {
 592                 errorOut(e, "PCACommand", "print_matrix");
 593                 exit(1);
 594         }
 595 }
 596 /*********************************************************************************************************************************/
 597