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 = 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 += 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;
101 //if(fbase.find_last_of(".")!=string::npos){
102 // fbase.erase(fbase.find_last_of(".")+1);
108 fbase = outputDir + getRootName(getSimpleName(filename));
110 read(filename, names, D);
112 double offset = 0.0000;
115 vector<vector<double> > G = D;
116 vector<vector<double> > copy_G;
117 //int rank = D.size();
119 cout << "\nProcessing...\n";
121 for(int count=0;count<2;count++){
122 recenter(offset, D, G);
125 offset = d[d.size()-1];
126 if(offset > 0.0) break;
130 output(fbase, names, G, d);
132 m->mothurOutEndLine();
133 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
134 for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
135 m->mothurOutEndLine();
139 catch(exception& e) {
140 m->errorOut(e, "PCACommand", "execute");
144 /*********************************************************************************************************************************/
146 inline double SIGN(const double a, const double b)
148 return b>=0 ? (a>=0 ? a:-a) : (a>=0 ? -a:a);
151 /*********************************************************************************************************************************/
153 void PCACommand::get_comment(istream& f, char begin, char end){
156 while(d != end){ d = f.get(); }
159 catch(exception& e) {
160 m->errorOut(e, "PCACommand", "get_comment");
165 /*********************************************************************************************************************************/
167 void PCACommand::read_phylip(istream& f, int square_m, vector<string>& name_list, vector<vector<double> >& d){
175 name_list.resize(rank);
178 for(int i=0;i<rank;i++)
180 for(int i=0;i<rank;i++) {
182 // cout << i << "\t" << name_list[i] << endl;
183 for(int j=0;j<rank;j++) {
185 if (d[i][j] == -0.0000)
190 else if(square_m == 2){
191 for(int i=0;i<rank;i++){
196 for(int i=1;i<rank;i++){
199 for(int j=0;j<i;j++){
201 if (d[i][j] == -0.0000)
208 catch(exception& e) {
209 m->errorOut(e, "PCACommand", "read_phylip");
215 /*********************************************************************************************************************************/
217 void PCACommand::read(string fname, vector<string>& names, vector<vector<double> >& D){
220 openInputFile(fname, f);
222 //check whether matrix is square
228 f >> numSeqs >> name;
230 while((d=f.get()) != EOF){
232 //is d a number meaning its square
238 //is d a line return meaning its lower triangle
246 //reopen to get back to beginning
247 openInputFile(fname, f);
248 read_phylip(f, m, names, D);
250 catch(exception& e) {
251 m->errorOut(e, "PCACommand", "read");
256 /*********************************************************************************************************************************/
258 double PCACommand::pythag(double a, double b) { return(pow(a*a+b*b,0.5)); }
260 /*********************************************************************************************************************************/
262 void PCACommand::matrix_mult(vector<vector<double> > first, vector<vector<double> > second, vector<vector<double> >& product){
264 int first_rows = first.size();
265 int first_cols = first[0].size();
266 int second_cols = second[0].size();
268 product.resize(first_rows);
269 for(int i=0;i<first_rows;i++){
270 product[i].resize(second_cols);
273 for(int i=0;i<first_rows;i++){
274 for(int j=0;j<second_cols;j++){
276 for(int k=0;k<first_cols;k++){
277 product[i][j] += first[i][k] * second[k][j];
282 catch(exception& e) {
283 m->errorOut(e, "PCACommand", "matrix_mult");
289 /*********************************************************************************************************************************/
291 void PCACommand::recenter(double offset, vector<vector<double> > D, vector<vector<double> >& G){
295 vector<vector<double> > A(rank);
296 vector<vector<double> > C(rank);
297 for(int i=0;i<rank;i++){
302 double scale = -1.0000 / (double) rank;
304 for(int i=0;i<rank;i++){
306 C[i][i] = 1.0000 + scale;
307 for(int j=i+1;j<rank;j++){
308 A[i][j] = A[j][i] = -0.5 * D[i][j] * D[i][j] + offset;
309 C[i][j] = C[j][i] = scale;
316 catch(exception& e) {
317 m->errorOut(e, "PCACommand", "recenter");
323 /*********************************************************************************************************************************/
325 // This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
327 void PCACommand::tred2(vector<vector<double> >& a, vector<double>& d, vector<double>& e){
329 double scale, hh, h, g, f;
336 for(int i=n-1;i>0;i--){
340 for(int k=0;k<l+1;k++){
341 scale += fabs(a[i][k]);
347 for(int k=0;k<l+1;k++){
349 h += a[i][k] * a[i][k];
352 g = (f >= 0.0 ? -sqrt(h) : sqrt(h));
357 for(int j=0;j<l+1;j++){
358 a[j][i] = a[i][j] / h;
360 for(int k=0;k<j+1;k++){
361 g += a[j][k] * a[i][k];
363 for(int k=j+1;k<l+1;k++){
364 g += a[k][j] * a[i][k];
370 for(int j=0;j<l+1;j++){
372 e[j] = g = e[j] - hh * f;
373 for(int k=0;k<j+1;k++){
374 a[j][k] -= (f * e[k] + g * a[i][k]);
389 for(int i=0;i<n;i++){
392 for(int j=0;j<l;j++){
394 for(int k=0;k<l;k++){
395 g += a[i][k] * a[k][j];
397 for(int k=0;k<l;k++){
398 a[k][j] -= g * a[k][i];
404 for(int j=0;j<l;j++){
405 a[j][i] = a[i][j] = 0.0;
409 catch(exception& e) {
410 m->errorOut(e, "PCACommand", "tred2");
416 /*********************************************************************************************************************************/
418 // This function is taken from Numerical Recipes in C++ by Press et al., 2nd edition, pg. 479
420 void PCACommand::qtli(vector<double>& d, vector<double>& e, vector<vector<double> >& z) {
423 double s, r, p, g, f, dd, c, b;
426 for(int i=1;i<=n;i++){
431 for(int l=0;l<n;l++){
435 dd = fabs(d[m]) + fabs(d[m+1]);
436 if(fabs(e[m])+dd == dd) break;
439 if(iter++ == 30) cerr << "Too many iterations in tqli\n";
440 g = (d[l+1]-d[l]) / (2.0 * e[l]);
442 g = d[m] - d[l] + e[l] / (g + SIGN(r,g));
448 e[i+1] = (r=pythag(f,g));
457 r = (d[i] - g) * s + 2.0 * c * b;
458 d[i+1] = g + ( p = s * r);
460 for(int k=0;k<n;k++){
462 z[k][i+1] = s * z[k][i] + c * f;
463 z[k][i] = c * z[k][i] - s * f;
466 if(r == 0.00 && i >= l) continue;
475 for(int i=0;i<n;i++){
477 for(int j=i;j<n;j++){
485 for(int j=0;j<n;j++){
493 catch(exception& e) {
494 m->errorOut(e, "PCACommand", "qtli");
499 /*********************************************************************************************************************************/
501 void PCACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> > G, vector<double> d) {
503 int rank = name_list.size();
504 double dsum = 0.0000;
505 for(int i=0;i<rank;i++){
507 for(int j=0;j<rank;j++){
508 if(d[j] >= 0) { G[i][j] *= pow(d[j],0.5); }
509 else { G[i][j] = 0.00000; }
513 ofstream pcaData((fnameRoot+"pca").c_str(), ios::trunc);
514 pcaData.setf(ios::fixed, ios::floatfield);
515 pcaData.setf(ios::showpoint);
516 outputNames.push_back(fnameRoot+"pca");
518 ofstream pcaLoadings((fnameRoot+"pca.loadings").c_str(), ios::trunc);
519 pcaLoadings.setf(ios::fixed, ios::floatfield);
520 pcaLoadings.setf(ios::showpoint);
521 outputNames.push_back(fnameRoot+"pca.loadings");
523 pcaLoadings << "axis\tloading\n";
524 for(int i=0;i<rank;i++){
525 pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
529 for(int i=0;i<rank;i++){
530 pcaData << '\t' << "axis" << i+1;
534 for(int i=0;i<rank;i++){
535 pcaData << name_list[i] << '\t';
536 for(int j=0;j<rank;j++){
537 pcaData << G[i][j] << '\t';
542 catch(exception& e) {
543 m->errorOut(e, "PCACommand", "output");
548 /*********************************************************************************************************************************/