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","outputdir", "inputdir"};
  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                         map<string, string>::iterator it;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37                         //if the user changes the input directory command factory will send this info to us in the output parameter
  38                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  39                         if (inputDir == "not found"){   inputDir = "";          }
  40                         else {
  41                                 string path;
  42                                 it = parameters.find("phylip");
  43                                 //user has given a template file
  44                                 if(it != parameters.end()){
  45                                         path = hasPath(it->second);
  46                                         //if the user has not given a path then, add inputdir. else leave path alone.
  47                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  48                                 }
  49                         }
  50
  51                         //required parameters
  52                         phylipfile = validParameter.validFile(parameters, "phylip", true);
  53                         if (phylipfile == "not open") { abort = true; }
  54                         else if (phylipfile == "not found") { phylipfile = ""; abort = true; }
  55                         else {  filename = phylipfile;  }
  56
  57                         //if the user changes the output directory command factory will send this info to us in the output parameter
  58                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  59                                 outputDir = "";
  60                                 outputDir += hasPath(phylipfile); //if user entered a file with a path then preserve it
  61                         }
  62
  63                         //error checking on files
  64                         if (phylipfile == "")   { mothurOut("You must provide a distance file before running the pca command."); mothurOutEndLine(); abort = true; }
  65                 }
  66
  67         }
  68         catch(exception& e) {
  69                 errorOut(e, "PCACommand", "PCACommand");
  70                 exit(1);
  71         }
  72 }
  73 //**********************************************************************************************************************
  74 void PCACommand::help(){
  75         try {
  76
  77                 mothurOut("The pca command..."); mothurOutEndLine();
  78         }
  79         catch(exception& e) {
  80                 errorOut(e, "PCACommand", "help");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 PCACommand::~PCACommand(){}
  86 //**********************************************************************************************************************
  87 int PCACommand::execute(){
  88         try {
  89
  90                 if (abort == true) { return 0; }
  91
  92                 cout.setf(ios::fixed, ios::floatfield);
  93                 cout.setf(ios::showpoint);
  94                 cerr.setf(ios::fixed, ios::floatfield);
  95                 cerr.setf(ios::showpoint);
  96
  97                 vector<string> names;
  98                 vector<vector<double> > D;
  99
 100                 //fbase = filename;
 101                 //if(fbase.find_last_of(".")!=string::npos){
 102                 //      fbase.erase(fbase.find_last_of(".")+1);
 103                 //}
 104                 //else{
 105                 //      fbase += ".";
 106                 //}
 107
 108                 fbase = outputDir + getRootName(getSimpleName(filename));
 109
 110                 read(filename, names, D);
 111
 112                 double offset = 0.0000;
 113                 vector<double> d;
 114                 vector<double> e;
 115                 vector<vector<double> > G = D;
 116                 vector<vector<double> > copy_G;
 117                 //int rank = D.size();
 118
 119                 cout << "\nProcessing...\n";
 120
 121                 for(int count=0;count<2;count++){
 122                         recenter(offset, D, G);
 123                         tred2(G, d, e);
 124                         qtli(d, e, G);
 125                         offset = d[d.size()-1];
 126                         if(offset > 0.0) break;
 127                 }
 128
 129
 130                 output(fbase, names, G, d);
 131
 132                 return 0;
 133         }
 134         catch(exception& e) {
 135                 errorOut(e, "PCACommand", "execute");
 136                 exit(1);
 137         }
 138 }
 139 /*********************************************************************************************************************************/
 140
 141 inline double SIGN(const double a, const double b)
 142 {
 143     return b>=0 ? (a>=0 ? a:-a) : (a>=0 ? -a:a);
 144 }
 145
 146 /*********************************************************************************************************************************/
 147
 148 void PCACommand::get_comment(istream& f, char begin, char end){
 149         try {
 150                 char d=f.get();
 151                 while(d != end){        d = f.get();    }
 152                 d = f.peek();
 153         }
 154         catch(exception& e) {
 155                 errorOut(e, "PCACommand", "get_comment");
 156                 exit(1);
 157         }
 158 }
 159
 160 /*********************************************************************************************************************************/
 161
 162 void PCACommand::read_mega(istream& f, vector<string>& name_list, vector<vector<double> >& d){
 163         try {
 164                 get_comment(f, '#', '\n');
 165
 166                 char test = f.peek();
 167
 168                 while(test == '!'){                                                             //get header comments
 169                         get_comment(f, '!', ';');
 170                         while(isspace(test=f.get()))            {;}
 171                         f.putback(test);
 172                         test = f.peek();
 173                 }
 174                 while(test != '\n'){                                                    //get sequence names
 175                         get_comment(f, '[', ']');
 176                         char d = f.get();
 177                         d = f.get();
 178                         if(d == '#'){
 179                                 string name;
 180                                 f >> name;
 181                                 name_list.push_back(name);
 182                                 while(isspace(test=f.get()))            {;}
 183                                 f.putback(test);
 184                         }
 185                         else{
 186                                 break;
 187                         }
 188                 }
 189                 int rank = name_list.size();
 190                 d.resize(rank);
 191                 for(int i=0;i<rank;i++){                d[i].resize(rank);              }
 192
 193                 d[0][0] = 0.0000;
 194                 get_comment(f, '[', ']');
 195                 for(int i=1;i<rank;i++){
 196                         get_comment(f, '[', ']');
 197                         d[i][i]=0.0000;
 198                         for(int j=0;j<i;j++){
 199                                 f >> d[i][j];
 200                                 if (d[i][j] == -0.0000)
 201                                         d[i][j] = 0.0000;
 202                                 d[j][i]=d[i][j];
 203                         }
 204                 }
 205         }
 206         catch(exception& e) {
 207                 errorOut(e, "PCACommand", "read_mega");
 208                 exit(1);
 209         }
 210 }
 211
 212 /*********************************************************************************************************************************/
 213
 214 void PCACommand::read_phylip(istream& f, int square_m, vector<string>& name_list, vector<vector<double> >& d){
 215         try {
 216                 //     int count1=0;
 217                 //     int count2=0;
 218
 219                 int rank;
 220                 f >> rank;
 221
 222                 name_list.resize(rank);
 223                 d.resize(rank);
 224                 if(square_m == 1){
 225                         for(int i=0;i<rank;i++)
 226                                 d[i].resize(rank);
 227                         for(int i=0;i<rank;i++) {
 228                                 f >> name_list[i];
 229                                 //                      cout << i << "\t" << name_list[i] << endl;
 230                                 for(int j=0;j<rank;j++) {
 231                                         f >> d[i][j];
 232                                         if (d[i][j] == -0.0000)
 233                                                 d[i][j] = 0.0000;
 234                                 }
 235                         }
 236                 }
 237                 else if(square_m == 2){
 238                         for(int i=0;i<rank;i++){
 239                                 d[i].resize(rank);
 240                         }
 241                         d[0][0] = 0.0000;
 242                         f >> name_list[0];
 243                         for(int i=1;i<rank;i++){
 244                                 f >> name_list[i];
 245                                 d[i][i]=0.0000;
 246                                 for(int j=0;j<i;j++){
 247                                         f >> d[i][j];
 248                                         if (d[i][j] == -0.0000)
 249                                                 d[i][j] = 0.0000;
 250                                         d[j][i]=d[i][j];
 251                                 }
 252                         }
 253                 }
 254         }
 255         catch(exception& e) {
 256                 errorOut(e, "PCACommand", "read_phylip");
 257                 exit(1);
 258         }
 259
 260 }
 261
 262 /*********************************************************************************************************************************/
 263
 264 void PCACommand::read(string fname, vector<string>& names, vector<vector<double> >& D){
 265         try {
 266                 ifstream f;
 267                 openInputFile(fname, f);
 268
 269                 char test = f.peek();
 270
 271                 if(test == '#'){
 272                         read_mega(f, names, D);
 273                 }
 274                 else{
 275                         //check whether matrix is square
 276                         char d;
 277                         int m = 1;
 278                         int numSeqs;
 279                         string name;
 280
 281                         f >> numSeqs >> name;
 282
 283                         while((d=f.get()) != EOF){
 284
 285                                 //is d a number meaning its square
 286                                 if(isalnum(d)){
 287                                         m = 1;
 288                                         break;
 289                                 }
 290
 291                                 //is d a line return meaning its lower triangle
 292                                 if(d == '\n'){
 293                                         m = 2;
 294                                         break;
 295                                 }
 296                         }
 297                         f.close();
 298
 299                         //reopen to get back to beginning
 300                         openInputFile(fname, f);
 301                         read_phylip(f, m, names, D);
 302                 }
 303
 304                 //int rank = D.size();
 305         }
 306         catch(exception& e) {
 307                 errorOut(e, "PCACommand", "read");
 308                 exit(1);
 309         }
 310 }
 311
 312 /*********************************************************************************************************************************/
 313 double PCACommand::pythag(double a, double b){
 314     return(pow(a*a+b*b,0.5));
 315 }
 316 /*********************************************************************************************************************************/
 317
 318 void PCACommand::matrix_mult(vector<vector<double> > first, vector<vector<double> > second, vector<vector<double> >& product){
 319         try {
 320                 int first_rows = first.size();
 321                 int first_cols = first[0].size();
 322                 int second_cols = second[0].size();
 323
 324                 product.resize(first_rows);
 325                 for(int i=0;i<first_rows;i++){
 326                         product[i].resize(second_cols);
 327                 }
 328
 329                 for(int i=0;i<first_rows;i++){
 330                         for(int j=0;j<second_cols;j++){
 331                                 product[i][j] = 0.0;
 332                                 for(int k=0;k<first_cols;k++){
 333                                         product[i][j] += first[i][k] * second[k][j];
 334                                 }
 335                         }
 336                 }
 337         }
 338         catch(exception& e) {
 339                 errorOut(e, "PCACommand", "matrix_mult");
 340                 exit(1);
 341         }
 342
 343 }
 344
 345 /*********************************************************************************************************************************/
 346
 347 void PCACommand::recenter(double offset, vector<vector<double> > D, vector<vector<double> >& G){
 348         try {
 349                 int rank = D.size();
 350
 351                 vector<vector<double> > A(rank);
 352                 vector<vector<double> > C(rank);
 353                 for(int i=0;i<rank;i++){
 354                         A[i].resize(rank);
 355                         C[i].resize(rank);
 356                 }
 357
 358                 double scale = -1.0000 / (double) rank;
 359
 360                 for(int i=0;i<rank;i++){
 361                         A[i][i] = 0.0000;
 362                         C[i][i] = 1.0000 + scale;
 363                         for(int j=i+1;j<rank;j++){
 364                                 A[i][j] = A[j][i] = -0.5 * D[i][j] * D[i][j] + offset;
 365                                 C[i][j] = C[j][i] = scale;
 366                         }
 367                 }
 368
 369                 matrix_mult(C,A,A);
 370                 matrix_mult(A,C,G);
 371         }
 372         catch(exception& e) {
 373                 errorOut(e, "PCACommand", "recenter");
 374                 exit(1);
 375         }
 376
 377 }
 378
 379 /*********************************************************************************************************************************/
 380
 381 //  This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
 382
 383 void PCACommand::tred2(vector<vector<double> >& a, vector<double>& d, vector<double>& e){
 384         try {
 385                 double scale, hh, h, g, f;
 386
 387                 int n = a.size();
 388
 389                 d.resize(n);
 390                 e.resize(n);
 391
 392                 for(int i=n-1;i>0;i--){
 393                         int l=i-1;
 394                         h = scale = 0.0000;
 395                         if(l>0){
 396                                 for(int k=0;k<l+1;k++){
 397                                         scale += fabs(a[i][k]);
 398                                 }
 399                                 if(scale == 0.0){
 400                                         e[i] = a[i][l];
 401                                 }
 402                                 else{
 403                                         for(int k=0;k<l+1;k++){
 404                                                 a[i][k] /= scale;
 405                                                 h += a[i][k] * a[i][k];
 406                                         }
 407                                         f = a[i][l];
 408                                         g = (f >= 0.0 ? -sqrt(h) : sqrt(h));
 409                                         e[i] = scale * g;
 410                                         h -= f * g;
 411                                         a[i][l] = f - g;
 412                                         f = 0.0;
 413                                         for(int j=0;j<l+1;j++){
 414                                                 a[j][i] = a[i][j] / h;
 415                                                 g = 0.0;
 416                                                 for(int k=0;k<j+1;k++){
 417                                                         g += a[j][k] * a[i][k];
 418                                                 }
 419                                                 for(int k=j+1;k<l+1;k++){
 420                                                         g += a[k][j] * a[i][k];
 421                                                 }
 422                                                 e[j] = g / h;
 423                                                 f += e[j] * a[i][j];
 424                                         }
 425                                         hh = f / (h + h);
 426                                         for(int j=0;j<l+1;j++){
 427                                                 f = a[i][j];
 428                                                 e[j] = g = e[j] - hh * f;
 429                                                 for(int k=0;k<j+1;k++){
 430                                                         a[j][k] -= (f * e[k] + g * a[i][k]);
 431                                                 }
 432                                         }
 433                                 }
 434                         }
 435                         else{
 436                                 e[i] = a[i][l];
 437                         }
 438
 439                         d[i] = h;
 440                 }
 441
 442                 d[0] = 0.0000;
 443                 e[0] = 0.0000;
 444
 445                 for(int i=0;i<n;i++){
 446                         int l = i;
 447                         if(d[i] != 0.0){
 448                                 for(int j=0;j<l;j++){
 449                                         g = 0.0000;
 450                                         for(int k=0;k<l;k++){
 451                                                 g += a[i][k] * a[k][j];
 452                                         }
 453                                         for(int k=0;k<l;k++){
 454                                                 a[k][j] -= g * a[k][i];
 455                                         }
 456                                 }
 457                         }
 458                         d[i] = a[i][i];
 459                         a[i][i] = 1.0000;
 460                         for(int j=0;j<l;j++){
 461                                 a[j][i] = a[i][j] = 0.0;
 462                         }
 463                 }
 464         }
 465         catch(exception& e) {
 466                 errorOut(e, "PCACommand", "tred2");
 467                 exit(1);
 468         }
 469
 470 }
 471
 472 /*********************************************************************************************************************************/
 473
 474 //  This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
 475
 476 void PCACommand::qtli(vector<double>& d, vector<double>& e, vector<vector<double> >& z) {
 477         try {
 478                 int m, i, iter;
 479                 double s, r, p, g, f, dd, c, b;
 480
 481                 int n = d.size();
 482                 for(int i=1;i<=n;i++){
 483                         e[i-1] = e[i];
 484                 }
 485                 e[n-1] = 0.0000;
 486
 487                 for(int l=0;l<n;l++){
 488                         iter = 0;
 489                         do {
 490                                 for(m=l;m<n-1;m++){
 491                                         dd = fabs(d[m]) + fabs(d[m+1]);
 492                                         if(fabs(e[m])+dd == dd) break;
 493                                 }
 494                                 if(m != l){
 495                                         if(iter++ == 30) cerr << "Too many iterations in tqli\n";
 496                                         g = (d[l+1]-d[l]) / (2.0 * e[l]);
 497                                         r = pythag(g, 1.0);
 498                                         g = d[m] - d[l] + e[l] / (g + SIGN(r,g));
 499                                         s = c = 1.0;
 500                                         p = 0.0000;
 501                                         for(i=m-1;i>=l;i--){
 502                                                 f = s * e[i];
 503                                                 b = c * e[i];
 504                                                 e[i+1] = (r=pythag(f,g));
 505                                                 if(r==0.0){
 506                                                         d[i+1] -= p;
 507                                                         e[m] = 0.0000;
 508                                                         break;
 509                                                 }
 510                                                 s = f / r;
 511                                                 c = g / r;
 512                                                 g = d[i+1] - p;
 513                                                 r = (d[i] - g) * s + 2.0 * c * b;
 514                                                 d[i+1] = g + ( p = s * r);
 515                                                 g = c * r - b;
 516                                                 for(int k=0;k<n;k++){
 517                                                         f = z[k][i+1];
 518                                                         z[k][i+1] = s * z[k][i] + c * f;
 519                                                         z[k][i] = c * z[k][i] - s * f;
 520                                                 }
 521                                         }
 522                                         if(r == 0.00 && i >= l) continue;
 523                                         d[l] -= p;
 524                                         e[l] = g;
 525                                         e[m] = 0.0;
 526                                 }
 527                         } while (m != l);
 528                 }
 529
 530                 int k;
 531                 for(int i=0;i<n;i++){
 532                         p=d[k=i];
 533                         for(int j=i;j<n;j++){
 534                                 if(d[j] >= p){
 535                                         p=d[k=j];
 536                                 }
 537                         }
 538                         if(k!=i){
 539                                 d[k]=d[i];
 540                                 d[i]=p;
 541                                 for(int j=0;j<n;j++){
 542                                         p=z[j][i];
 543                                         z[j][i] = z[j][k];
 544                                         z[j][k] = p;
 545                                 }
 546                         }
 547                 }
 548         }
 549         catch(exception& e) {
 550                 errorOut(e, "PCACommand", "qtli");
 551                 exit(1);
 552         }
 553 }
 554
 555 /*********************************************************************************************************************************/
 556
 557 void PCACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> > G, vector<double> d) {
 558         try {
 559                 int rank = name_list.size();
 560                 double dsum = 0.0000;
 561                 for(int i=0;i<rank;i++){
 562                         dsum += d[i];
 563                         for(int j=0;j<rank;j++){
 564                                 if(d[j] >= 0)   {       G[i][j] *= pow(d[j],0.5);       }
 565                                 else                    {       G[i][j] = 0.00000;                      }
 566                         }
 567                 }
 568
 569                 ofstream pcaData((fnameRoot+"pca").c_str(), ios::trunc);
 570                 pcaData.setf(ios::fixed, ios::floatfield);
 571                 pcaData.setf(ios::showpoint);
 572
 573                 ofstream pcaLoadings((fnameRoot+"pca.loadings").c_str(), ios::trunc);
 574                 pcaLoadings.setf(ios::fixed, ios::floatfield);
 575                 pcaLoadings.setf(ios::showpoint);
 576
 577                 pcaLoadings << "axis\tloading\n";
 578                 for(int i=0;i<rank;i++){
 579                         pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
 580                 }
 581
 582                 pcaData << "SeqName";
 583                 for(int i=0;i<rank;i++){
 584                         pcaData << '\t' << "axis" << i+1;
 585                 }
 586                 pcaData << endl;
 587
 588                 for(int i=0;i<rank;i++){
 589                         pcaData << name_list[i] << '\t';
 590                         for(int j=0;j<rank;j++){
 591                                 pcaData << G[i][j] << '\t';
 592                         }
 593                         pcaData << endl;
 594                 }
 595         }
 596         catch(exception& e) {
 597                 errorOut(e, "PCACommand", "output");
 598                 exit(1);
 599         }
 600 }
 601
 602 /*********************************************************************************************************************************/
 603
 604 void PCACommand::print_matrix(vector<vector<double> > A) {
 605         try {
 606                 int rank = A.size();
 607                 for(int i=0;i<rank;i++){
 608                         for(int j=0;j<rank;j++){
 609                                 cout << A[i][j] << "  ";
 610                         }
 611                         cout << endl;
 612                 }
 613         }
 614         catch(exception& e) {
 615                 errorOut(e, "PCACommand", "print_matrix");
 616                 exit(1);
 617         }
 618 }
 619 /*********************************************************************************************************************************/
 620