5 * Created by Sarah Westcott on 3/9/09.
6 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
10 /* This class is designed to implement an integral form of the Cramer-von Mises statistic.
11 you may refer to the "Integration of Microbial Ecology and Statistics: A Test To Compare Gene Libraries"
12 paper in Applied and Environmental Microbiology, Sept. 2004, p. 5485-5492 0099-2240/04/$8.00+0
13 DOI: 10.1128/AEM.70.9.5485-5492.2004 Copyright 2004 American Society for Microbiology for more information. */
16 #include "libshuffcommand.h"
18 #include "slibshuff.h"
19 #include "dlibshuff.h"
21 //**********************************************************************************************************************
23 LibShuffCommand::LibShuffCommand(string option){
25 globaldata = GlobalData::getInstance();
30 //allow user to run help
31 if(option == "help") { help(); abort = true; }
34 //valid paramters for this command
35 string Array[] = {"iters","groups","step","form","cutoff"};
36 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
38 OptionParser parser(option);
39 map<string, string> parameters = parser.getParameters();
41 ValidParameters validParameter;
43 //check to make sure all parameters are valid for command
44 for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
45 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
48 //make sure the user has already run the read.dist command
49 if ((globaldata->gMatrix == NULL) || (globaldata->gGroupmap == NULL)) {
50 mothurOut("You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command. "); mothurOutEndLine(); abort = true;;
53 //check for optional parameter and set defaults
54 // ...at some point should added some additional type checking...
55 groups = validParameter.validFile(parameters, "groups", false);
56 if (groups == "not found") { groups = ""; savegroups = groups; }
59 splitAtDash(groups, Groups);
60 globaldata->Groups = Groups;
64 temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "10000"; }
67 temp = validParameter.validFile(parameters, "cutoff", false); if (temp == "not found") { temp = "1.0"; }
68 convert(temp, cutOff);
70 temp = validParameter.validFile(parameters, "step", false); if (temp == "not found") { temp = "0.01"; }
73 userform = validParameter.validFile(parameters, "form", false); if (userform == "not found") { userform = "integral"; }
77 matrix = globaldata->gMatrix; //get the distance matrix
78 setGroups(); //set the groups to be analyzed and sorts them
80 /********************************************************************************************/
81 //this is needed because when we read the matrix we sort it into groups in alphabetical order
82 //the rest of the command and the classes used in this command assume specific order
83 /********************************************************************************************/
84 matrix->setGroups(globaldata->gGroupmap->namesOfGroups);
86 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) { sizes.push_back(globaldata->gGroupmap->getNumSeqs(globaldata->gGroupmap->namesOfGroups[i])); }
87 matrix->setSizes(sizes);
90 if(userform == "discrete"){
91 form = new DLibshuff(matrix, iters, step, cutOff);
94 form = new SLibshuff(matrix, iters, cutOff);
101 catch(exception& e) {
102 errorOut(e, "LibShuffCommand", "LibShuffCommand");
106 //**********************************************************************************************************************
108 void LibShuffCommand::help(){
110 mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
111 mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff. No parameters are required.\n");
112 mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed. You must enter at least 2 valid groups.\n");
113 mothurOut("The group names are separated by dashes. The iters parameter allows you to specify how many random matrices you would like compared to your matrix.\n");
114 mothurOut("The step parameter allows you to specify change in distance you would like between each output if you are using the discrete form.\n");
115 mothurOut("The form parameter allows you to specify if you would like to analyze your matrix using the discrete or integral form. Your options are integral or discrete.\n");
116 mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
117 mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
118 mothurOut("The default value for groups is all the groups in your groupfile, iters is 10000, cutoff is 1.0, form is integral and step is 0.01.\n");
119 mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
120 mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
122 catch(exception& e) {
123 errorOut(e, "LibShuffCommand", "help");
128 //**********************************************************************************************************************
130 int LibShuffCommand::execute(){
133 if (abort == true) { return 0; }
135 savedDXYValues = form->evaluateAll();
136 savedMinValues = form->getSavedMins();
138 pValueCounts.resize(numGroups);
139 for(int i=0;i<numGroups;i++){
140 pValueCounts[i].assign(numGroups, 0);
143 Progress* reading = new Progress();
145 for(int i=0;i<numGroups-1;i++) {
146 for(int j=i+1;j<numGroups;j++) {
147 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
148 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
149 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
151 for(int p=0;p<iters;p++) {
152 form->randomizeGroups(spoti,spotj);
153 if(form->evaluatePair(spoti,spotj) >= savedDXYValues[spoti][spotj]) { pValueCounts[i][j]++; }
154 if(form->evaluatePair(spotj,spoti) >= savedDXYValues[spotj][spoti]) { pValueCounts[j][i]++; }
157 form->resetGroup(spoti);
158 form->resetGroup(spotj);
168 //clear out users groups
169 globaldata->Groups.clear();
172 //delete globaldata's copy of the gmatrix to free up memory
173 delete globaldata->gMatrix; globaldata->gMatrix = NULL;
177 catch(exception& e) {
178 errorOut(e, "LibShuffCommand", "execute");
183 //**********************************************************************************************************************
185 void LibShuffCommand::printCoverageFile() {
189 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.coverage";
190 openOutputFile(summaryFile, outCov);
191 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
192 //cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
194 map<double,vector<int> > allDistances;
195 map<double,vector<int> >::iterator it;
197 vector<vector<int> > indices(numGroups);
198 int numIndices = numGroups * numGroups;
201 for(int i=0;i<numGroups;i++){
202 indices[i].assign(numGroups,0);
203 for(int j=0;j<numGroups;j++){
204 indices[i][j] = index++;
206 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
207 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
209 for(int k=0;k<savedMinValues[spoti][spotj].size();k++){
210 if(allDistances[savedMinValues[spoti][spotj][k]].size() != 0){
211 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]]++;
214 allDistances[savedMinValues[spoti][spotj][k]].assign(numIndices, 0);
215 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]] = 1;
220 it=allDistances.begin();
222 //cout << setprecision(8);
224 vector<int> prevRow = it->second;
227 for(;it!=allDistances.end();it++){
228 for(int i=0;i<it->second.size();i++){
229 it->second[i] += prevRow[i];
231 prevRow = it->second;
234 vector<int> lastRow = allDistances.rbegin()->second;
235 outCov << setprecision(8);
238 for (int i = 0; i < numGroups; i++){
239 outCov << '\t' << groupNames[i];
241 for (int i=0;i<numGroups;i++){
242 for(int j=i+1;j<numGroups;j++){
243 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
244 outCov << groupNames[j] << '-' << groupNames[i];
249 for(it=allDistances.begin();it!=allDistances.end();it++){
250 outCov << it->first << '\t';
251 for(int i=0;i<numGroups;i++){
252 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
254 for(int i=0;i<numGroups;i++){
255 for(int j=i+1;j<numGroups;j++){
256 outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
257 outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
264 catch(exception& e) {
265 errorOut(e, "LibShuffCommand", "printCoverageFile");
270 //**********************************************************************************************************************
272 void LibShuffCommand::printSummaryFile() {
276 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.summary";
277 openOutputFile(summaryFile, outSum);
279 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
280 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
282 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
283 mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); mothurOutEndLine();
284 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
286 int precision = (int)log10(iters);
287 for(int i=0;i<numGroups;i++){
288 for(int j=i+1;j<numGroups;j++){
289 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
290 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
292 if(pValueCounts[i][j]){
293 cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
294 mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); mothurOutEndLine();
295 outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
298 cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
299 mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
300 outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
302 if(pValueCounts[j][i]){
303 cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
304 mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); mothurOutEndLine();
305 outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
308 cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
309 mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
310 outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
317 catch(exception& e) {
318 errorOut(e, "LibShuffCommand", "printSummaryFile");
323 //**********************************************************************************************************************
325 void LibShuffCommand::setGroups() {
327 //if the user has not entered specific groups to analyze then do them all
328 if (globaldata->Groups.size() == 0) {
329 numGroups = globaldata->gGroupmap->getNumGroups();
330 for (int i=0; i < numGroups; i++) {
331 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
334 if (savegroups != "all") {
335 //check that groups are valid
336 for (int i = 0; i < globaldata->Groups.size(); i++) {
337 if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
338 mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); mothurOutEndLine();
339 // erase the invalid group from globaldata->Groups
340 globaldata->Groups.erase(globaldata->Groups.begin()+i);
344 //if the user only entered invalid groups
345 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
346 numGroups = globaldata->gGroupmap->getNumGroups();
347 for (int i=0; i < numGroups; i++) {
348 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
350 mothurOut("When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile."); mothurOutEndLine();
351 } else { numGroups = globaldata->Groups.size(); }
352 } else { //users wants all groups
353 numGroups = globaldata->gGroupmap->getNumGroups();
354 globaldata->Groups.clear();
355 for (int i=0; i < numGroups; i++) {
356 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
361 //sort so labels match
362 sort(globaldata->Groups.begin(), globaldata->Groups.end());
365 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
367 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) { globaldata->gGroupmap->groupIndex[globaldata->gGroupmap->namesOfGroups[i]] = i; }
369 groupNames = globaldata->Groups;
372 catch(exception& e) {
373 errorOut(e, "LibShuffCommand", "setGroups");
378 /***********************************************************/