6 * Created by westcott on 1/4/10.
7 * Copyright 2010 Schloss Lab. All rights reserved.
11 #include "pcacommand.h"
13 //**********************************************************************************************************************
15 PCACommand::PCACommand(string option) {
19 //allow user to run help
20 if(option == "help") { help(); abort = true; }
23 //valid paramters for this command
24 string Array[] = {"phylip","outputdir", "inputdir"};
25 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
27 OptionParser parser(option);
28 map<string, string> parameters = parser. getParameters();
30 ValidParameters validParameter;
31 map<string, string>::iterator it;
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; }
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 = ""; }
42 it = parameters.find("phylip");
43 //user has given a template file
44 if(it != parameters.end()){
45 path = m->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; }
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; }
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"){
60 outputDir += m->hasPath(phylipfile); //if user entered a file with a path then preserve it
63 //error checking on files
64 if (phylipfile == "") { m->mothurOut("You must provide a distance file before running the pca command."); m->mothurOutEndLine(); abort = true; }
69 m->errorOut(e, "PCACommand", "PCACommand");
73 //**********************************************************************************************************************
74 void PCACommand::help(){
77 m->mothurOut("The pca command..."); m->mothurOutEndLine();
80 m->errorOut(e, "PCACommand", "help");
84 //**********************************************************************************************************************
85 PCACommand::~PCACommand(){}
86 //**********************************************************************************************************************
87 int PCACommand::execute(){
90 if (abort == true) { return 0; }
92 cout.setf(ios::fixed, ios::floatfield);
93 cout.setf(ios::showpoint);
94 cerr.setf(ios::fixed, ios::floatfield);
95 cerr.setf(ios::showpoint);
98 vector<vector<double> > D;
100 fbase = outputDir + m->getRootName(m->getSimpleName(filename));
102 read(filename, names, D);
104 if (m->control_pressed) { return 0; }
106 double offset = 0.0000;
109 vector<vector<double> > G = D;
110 vector<vector<double> > copy_G;
111 //int rank = D.size();
113 m->mothurOut("\nProcessing...\n");
115 for(int count=0;count<2;count++){
116 recenter(offset, D, G); if (m->control_pressed) { return 0; }
117 tred2(G, d, e); if (m->control_pressed) { return 0; }
118 qtli(d, e, G); if (m->control_pressed) { return 0; }
119 offset = d[d.size()-1];
120 if(offset > 0.0) break;
123 if (m->control_pressed) { return 0; }
125 output(fbase, names, G, d);
127 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
129 m->mothurOutEndLine();
130 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
131 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
132 m->mothurOutEndLine();
136 catch(exception& e) {
137 m->errorOut(e, "PCACommand", "execute");
141 /*********************************************************************************************************************************/
143 inline double SIGN(const double a, const double b)
145 return b>=0 ? (a>=0 ? a:-a) : (a>=0 ? -a:a);
148 /*********************************************************************************************************************************/
150 void PCACommand::get_comment(istream& f, char begin, char end){
153 while(d != end){ d = f.get(); }
156 catch(exception& e) {
157 m->errorOut(e, "PCACommand", "get_comment");
162 /*********************************************************************************************************************************/
164 int PCACommand::read_phylip(istream& f, int square_m, vector<string>& name_list, vector<vector<double> >& d){
172 name_list.resize(rank);
175 for(int i=0;i<rank;i++)
177 for(int i=0;i<rank;i++) {
179 // cout << i << "\t" << name_list[i] << endl;
180 for(int j=0;j<rank;j++) {
181 if (m->control_pressed) { return 0; }
184 if (d[i][j] == -0.0000)
189 else if(square_m == 2){
190 for(int i=0;i<rank;i++){
195 for(int i=1;i<rank;i++){
198 for(int j=0;j<i;j++){
199 if (m->control_pressed) { return 0; }
201 if (d[i][j] == -0.0000)
210 catch(exception& e) {
211 m->errorOut(e, "PCACommand", "read_phylip");
217 /*********************************************************************************************************************************/
219 void PCACommand::read(string fname, vector<string>& names, vector<vector<double> >& D){
222 m->openInputFile(fname, f);
224 //check whether matrix is square
230 f >> numSeqs >> name;
232 while((d=f.get()) != EOF){
234 //is d a number meaning its square
240 //is d a line return meaning its lower triangle
248 //reopen to get back to beginning
249 m->openInputFile(fname, f);
250 read_phylip(f, q, names, D);
252 catch(exception& e) {
253 m->errorOut(e, "PCACommand", "read");
258 /*********************************************************************************************************************************/
260 double PCACommand::pythag(double a, double b) { return(pow(a*a+b*b,0.5)); }
262 /*********************************************************************************************************************************/
264 void PCACommand::matrix_mult(vector<vector<double> > first, vector<vector<double> > second, vector<vector<double> >& product){
266 int first_rows = first.size();
267 int first_cols = first[0].size();
268 int second_cols = second[0].size();
270 product.resize(first_rows);
271 for(int i=0;i<first_rows;i++){
272 product[i].resize(second_cols);
275 for(int i=0;i<first_rows;i++){
276 for(int j=0;j<second_cols;j++){
278 for(int k=0;k<first_cols;k++){
279 product[i][j] += first[i][k] * second[k][j];
284 catch(exception& e) {
285 m->errorOut(e, "PCACommand", "matrix_mult");
291 /*********************************************************************************************************************************/
293 void PCACommand::recenter(double offset, vector<vector<double> > D, vector<vector<double> >& G){
297 vector<vector<double> > A(rank);
298 vector<vector<double> > C(rank);
299 for(int i=0;i<rank;i++){
304 double scale = -1.0000 / (double) rank;
306 for(int i=0;i<rank;i++){
308 C[i][i] = 1.0000 + scale;
309 for(int j=i+1;j<rank;j++){
310 A[i][j] = A[j][i] = -0.5 * D[i][j] * D[i][j] + offset;
311 C[i][j] = C[j][i] = scale;
318 catch(exception& e) {
319 m->errorOut(e, "PCACommand", "recenter");
325 /*********************************************************************************************************************************/
327 // This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
329 void PCACommand::tred2(vector<vector<double> >& a, vector<double>& d, vector<double>& e){
331 double scale, hh, h, g, f;
338 for(int i=n-1;i>0;i--){
342 for(int k=0;k<l+1;k++){
343 scale += fabs(a[i][k]);
349 for(int k=0;k<l+1;k++){
351 h += a[i][k] * a[i][k];
354 g = (f >= 0.0 ? -sqrt(h) : sqrt(h));
359 for(int j=0;j<l+1;j++){
360 a[j][i] = a[i][j] / h;
362 for(int k=0;k<j+1;k++){
363 g += a[j][k] * a[i][k];
365 for(int k=j+1;k<l+1;k++){
366 g += a[k][j] * a[i][k];
372 for(int j=0;j<l+1;j++){
374 e[j] = g = e[j] - hh * f;
375 for(int k=0;k<j+1;k++){
376 a[j][k] -= (f * e[k] + g * a[i][k]);
391 for(int i=0;i<n;i++){
394 for(int j=0;j<l;j++){
396 for(int k=0;k<l;k++){
397 g += a[i][k] * a[k][j];
399 for(int k=0;k<l;k++){
400 a[k][j] -= g * a[k][i];
406 for(int j=0;j<l;j++){
407 a[j][i] = a[i][j] = 0.0;
411 catch(exception& e) {
412 m->errorOut(e, "PCACommand", "tred2");
418 /*********************************************************************************************************************************/
420 // This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
422 void PCACommand::qtli(vector<double>& d, vector<double>& e, vector<vector<double> >& z) {
425 double s, r, p, g, f, dd, c, b;
428 for(int i=1;i<=n;i++){
433 for(int l=0;l<n;l++){
437 dd = fabs(d[m]) + fabs(d[m+1]);
438 if(fabs(e[m])+dd == dd) break;
441 if(iter++ == 30) cerr << "Too many iterations in tqli\n";
442 g = (d[l+1]-d[l]) / (2.0 * e[l]);
444 g = d[m] - d[l] + e[l] / (g + SIGN(r,g));
450 e[i+1] = (r=pythag(f,g));
459 r = (d[i] - g) * s + 2.0 * c * b;
460 d[i+1] = g + ( p = s * r);
462 for(int k=0;k<n;k++){
464 z[k][i+1] = s * z[k][i] + c * f;
465 z[k][i] = c * z[k][i] - s * f;
468 if(r == 0.00 && i >= l) continue;
477 for(int i=0;i<n;i++){
479 for(int j=i;j<n;j++){
487 for(int j=0;j<n;j++){
495 catch(exception& e) {
496 m->errorOut(e, "PCACommand", "qtli");
501 /*********************************************************************************************************************************/
503 void PCACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> > G, vector<double> d) {
505 int rank = name_list.size();
506 double dsum = 0.0000;
507 for(int i=0;i<rank;i++){
509 for(int j=0;j<rank;j++){
510 if(d[j] >= 0) { G[i][j] *= pow(d[j],0.5); }
511 else { G[i][j] = 0.00000; }
515 ofstream pcaData((fnameRoot+"pcoa").c_str(), ios::trunc);
516 pcaData.setf(ios::fixed, ios::floatfield);
517 pcaData.setf(ios::showpoint);
518 outputNames.push_back(fnameRoot+"pcoa");
520 ofstream pcaLoadings((fnameRoot+"pcoa.loadings").c_str(), ios::trunc);
521 pcaLoadings.setf(ios::fixed, ios::floatfield);
522 pcaLoadings.setf(ios::showpoint);
523 outputNames.push_back(fnameRoot+"pcoa.loadings");
525 pcaLoadings << "axis\tloading\n";
526 for(int i=0;i<rank;i++){
527 pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
531 for(int i=0;i<rank;i++){
532 pcaData << '\t' << "axis" << i+1;
536 for(int i=0;i<rank;i++){
537 pcaData << name_list[i] << '\t';
538 for(int j=0;j<rank;j++){
539 pcaData << G[i][j] << '\t';
544 catch(exception& e) {
545 m->errorOut(e, "PCACommand", "output");
550 /*********************************************************************************************************************************/