#include "counttable.h"
+/************************************************************/
+int CountTable::createTable(set<string>& n, map<string, string>& g, set<string>& gs) {
+ try {
+ int numGroups = 0;
+ groups.clear();
+ totalGroups.clear();
+ indexGroupMap.clear();
+ indexNameMap.clear();
+ counts.clear();
+ for (set<string>::iterator it = gs.begin(); it != gs.end(); it++) { groups.push_back(*it); hasGroups = true; }
+ numGroups = groups.size();
+ totalGroups.resize(numGroups, 0);
+
+ //sort groups to keep consistent with how we store the groups in groupmap
+ sort(groups.begin(), groups.end());
+ for (int i = 0; i < groups.size(); i++) { indexGroupMap[groups[i]] = i; }
+ m->setAllGroups(groups);
+
+ uniques = 0;
+ total = 0;
+ for (set<string>::iterator it = n.begin(); it != n.end(); it++) {
+
+ if (m->control_pressed) { break; }
+
+ string seqName = *it;
+
+ vector<int> groupCounts; groupCounts.resize(numGroups, 0);
+ map<string, string>::iterator itGroup = g.find(seqName);
+
+ if (itGroup != g.end()) {
+ groupCounts[indexGroupMap[itGroup->second]] = 1;
+ totalGroups[indexGroupMap[itGroup->second]]++;
+ }else { m->mothurOut("[ERROR]: Your group file does not contain " + seqName + ". Please correct."); m->mothurOutEndLine(); }
+
+ map<string, int>::iterator it2 = indexNameMap.find(seqName);
+ if (it2 == indexNameMap.end()) {
+ if (hasGroups) { counts.push_back(groupCounts); }
+ indexNameMap[seqName] = uniques;
+ totals.push_back(1);
+ total++;
+ uniques++;
+ }
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "createTable");
+ exit(1);
+ }
+}
/************************************************************/
bool CountTable::testGroups(string file) {
try {
}
}
/************************************************************/
+int CountTable::createTable(string namefile, string groupfile, bool createGroup) {
+ try {
+
+ if (namefile == "") { m->mothurOut("[ERROR]: namefile cannot be blank when creating a count table.\n"); m->control_pressed = true; }
+
+ GroupMap* groupMap;
+ int numGroups = 0;
+ groups.clear();
+ totalGroups.clear();
+ indexGroupMap.clear();
+ indexNameMap.clear();
+ counts.clear();
+ map<int, string> originalGroupIndexes;
+
+ if (groupfile != "") {
+ hasGroups = true;
+ groupMap = new GroupMap(groupfile); groupMap->readMap();
+ numGroups = groupMap->getNumGroups();
+ groups = groupMap->getNamesOfGroups();
+ totalGroups.resize(numGroups, 0);
+ }else if(createGroup) {
+ hasGroups = true;
+ numGroups = 1;
+ groups.push_back("Group1");
+ totalGroups.resize(numGroups, 0);
+ }
+ //sort groups to keep consistent with how we store the groups in groupmap
+ sort(groups.begin(), groups.end());
+ for (int i = 0; i < groups.size(); i++) { indexGroupMap[groups[i]] = i; }
+ m->setAllGroups(groups);
+
+ bool error = false;
+ string name;
+ uniques = 0;
+ total = 0;
+
+
+ //open input file
+ ifstream in;
+ m->openInputFile(namefile, in);
+
+ int total = 0;
+ while (!in.eof()) {
+ if (m->control_pressed) { break; }
+
+ string firstCol, secondCol;
+ in >> firstCol; m->gobble(in); in >> secondCol; m->gobble(in);
+
+ vector<string> names;
+ m->splitAtChar(secondCol, names, ',');
+
+ map<string, int> groupCounts;
+ int thisTotal = 0;
+ if (groupfile != "") {
+ //set to 0
+ for (int i = 0; i < groups.size(); i++) { groupCounts[groups[i]] = 0; }
+
+ //get counts for each of the users groups
+ for (int i = 0; i < names.size(); i++) {
+ string group = groupMap->getGroup(names[i]);
+
+ if (group == "not found") { m->mothurOut("[ERROR]: " + names[i] + " is not in your groupfile, please correct."); m->mothurOutEndLine(); error=true; }
+ else {
+ map<string, int>::iterator it = groupCounts.find(group);
+
+ //if not found, then this sequence is not from a group we care about
+ if (it != groupCounts.end()) {
+ it->second++;
+ thisTotal++;
+ }
+ }
+ }
+ }else if (createGroup) {
+ groupCounts["Group1"]=0;
+ for (int i = 0; i < names.size(); i++) {
+ string group = "Group1";
+ groupCounts["Group1"]++; thisTotal++;
+ }
+ }else { thisTotal = names.size(); }
+
+ //if group info, then read it
+ vector<int> thisGroupsCount; thisGroupsCount.resize(numGroups, 0);
+ for (int i = 0; i < numGroups; i++) {
+ thisGroupsCount[i] = groupCounts[groups[i]];
+ totalGroups[i] += thisGroupsCount[i];
+ }
+
+ map<string, int>::iterator it = indexNameMap.find(firstCol);
+ if (it == indexNameMap.end()) {
+ if (hasGroups) { counts.push_back(thisGroupsCount); }
+ indexNameMap[firstCol] = uniques;
+ totals.push_back(thisTotal);
+ total += thisTotal;
+ uniques++;
+ }else {
+ error = true;
+ m->mothurOut("[ERROR]: Your count table contains more than 1 sequence named " + firstCol + ", sequence names must be unique. Please correct."); m->mothurOutEndLine();
+ }
+ }
+ in.close();
+
+ if (error) { m->control_pressed = true; }
+ if (groupfile != "") { delete groupMap; }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "createTable");
+ exit(1);
+ }
+}
+/************************************************************/
int CountTable::readTable(string file) {
try {
filename = file;
}
}
/************************************************************/
+int CountTable::printTable(string file) {
+ try {
+ ofstream out;
+ m->openOutputFile(file, out);
+ out << "Representative_Sequence\ttotal\t";
+ for (int i = 0; i < groups.size(); i++) { out << groups[i] << '\t'; }
+ out << endl;
+
+ for (map<string, int>::iterator itNames = indexNameMap.begin(); itNames != indexNameMap.end(); itNames++) {
+ out << itNames->first << '\t' << totals[itNames->second] << '\t';
+ if (hasGroups) {
+
+ for (int i = 0; i < groups.size(); i++) {
+ out << counts[itNames->second][i] << '\t';
+ }
+ }
+ out << endl;
+ }
+ out.close();
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "printTable");
+ exit(1);
+ }
+}
+/************************************************************/
+int CountTable::printHeaders(ofstream& out) {
+ try {
+ out << "Representative_Sequence\ttotal\t";
+ for (int i = 0; i < groups.size(); i++) { out << groups[i] << '\t'; }
+ out << endl;
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "printHeaders");
+ exit(1);
+ }
+}
+/************************************************************/
+int CountTable::printSeq(ofstream& out, string seqName) {
+ try {
+ map<string, int>::iterator it = indexNameMap.find(seqName);
+ if (it == indexNameMap.end()) {
+ m->mothurOut("[ERROR]: " + seqName + " is not in your count table. Please correct.\n"); m->control_pressed = true;
+ }else {
+ out << it->first << '\t' << totals[it->second] << '\t';
+ if (hasGroups) {
+ for (int i = 0; i < groups.size(); i++) {
+ out << counts[it->second][i] << '\t';
+ }
+ }
+ out << endl;
+ }
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "printSeq");
+ exit(1);
+ }
+}
+/************************************************************/
//group counts for a seq
vector<int> CountTable::getGroupCounts(string seqName) {
try {
exit(1);
}
}
+/************************************************************/
+//set the number of sequences for the seq for the group
+int CountTable::setAbund(string seqName, string groupName, int num) {
+ try {
+ if (hasGroups) {
+ map<string, int>::iterator it = indexGroupMap.find(groupName);
+ if (it == indexGroupMap.end()) {
+ m->mothurOut("[ERROR]: " + groupName + " is not in your count table. Please correct.\n"); m->control_pressed = true;
+ }else {
+ map<string, int>::iterator it2 = indexNameMap.find(seqName);
+ if (it2 == indexNameMap.end()) {
+ m->mothurOut("[ERROR]: " + seqName + " is not in your count table. Please correct.\n"); m->control_pressed = true;
+ }else {
+ int oldCount = counts[it2->second][it->second];
+ counts[it2->second][it->second] = num;
+ totalGroups[it->second] += (num - oldCount);
+ total += (num - oldCount);
+ totals[it2->second] += (num - oldCount);
+ }
+ }
+ }else{ m->mothurOut("[ERROR]: Your count table does not have group info. Please correct.\n"); m->control_pressed = true; }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "set");
+ exit(1);
+ }
+}
+/************************************************************/
+//add group
+int CountTable::addGroup(string groupName) {
+ try {
+ bool sanity = m->inUsersGroups(groupName, groups);
+ if (sanity) { m->mothurOut("[ERROR]: " + groupName + " is already in the count table, cannot add again.\n"); m->control_pressed = true; return 0; }
+
+ groups.push_back(groupName);
+ if (!hasGroups) { counts.resize(uniques); }
+
+ for (int i = 0; i < counts.size(); i++) { counts[i].push_back(0); }
+ totalGroups.push_back(0);
+ indexGroupMap[groupName] = groups.size()-1;
+ map<string, int> originalGroupMap = indexGroupMap;
+
+ //important to play well with others, :)
+ sort(groups.begin(), groups.end());
+
+ //fix indexGroupMap && totalGroups
+ vector<int> newTotals; newTotals.resize(groups.size(), 0);
+ for (int i = 0; i < groups.size(); i++) {
+ indexGroupMap[groups[i]] = i;
+ //find original spot of group[i]
+ int index = originalGroupMap[groups[i]];
+ newTotals[i] = totalGroups[index];
+ }
+ totalGroups = newTotals;
+
+ //fix counts vectors
+ for (int i = 0; i < counts.size(); i++) {
+ vector<int> newCounts; newCounts.resize(groups.size(), 0);
+ for (int j = 0; j < groups.size(); j++) {
+ //find original spot of group[i]
+ int index = originalGroupMap[groups[j]];
+ newCounts[j] = counts[i][index];
+ }
+ counts[i] = newCounts;
+ }
+ hasGroups = true;
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "addGroup");
+ exit(1);
+ }
+}
+/************************************************************/
+//vector of groups for the seq
+vector<string> CountTable::getGroups(string seqName) {
+ try {
+ vector<string> thisGroups;
+ if (hasGroups) {
+ vector<int> thisCounts = getGroupCounts(seqName);
+ for (int i = 0; i < thisCounts.size(); i++) {
+ if (thisCounts[i] != 0) { thisGroups.push_back(groups[i]); }
+ }
+ }else{ m->mothurOut("[ERROR]: Your count table does not have group info. Please correct.\n"); m->control_pressed = true; }
+
+ return thisGroups;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "getGroups");
+ exit(1);
+ }
+}
+/************************************************************/
+//total number of seqs represented by seq
+int CountTable::renameSeq(string oldSeqName, string newSeqName) {
+ try {
+
+ map<string, int>::iterator it = indexNameMap.find(oldSeqName);
+ if (it == indexNameMap.end()) {
+ m->mothurOut("[ERROR]: " + oldSeqName + " is not in your count table. Please correct.\n"); m->control_pressed = true;
+ }else {
+ int index = it->second;
+ indexNameMap.erase(it);
+ indexNameMap[newSeqName] = index;
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "renameSeq");
+ exit(1);
+ }
+}
+
/************************************************************/
//total number of seqs represented by seq
int CountTable::getNumSeqs(string seqName) {
}
}
/************************************************************/
+//remove sequence
+int CountTable::remove(string seqName) {
+ try {
+ map<string, int>::iterator it = indexNameMap.find(seqName);
+ if (it == indexNameMap.end()) {
+ uniques--;
+ if (hasGroups){ //remove this sequences counts from group totals
+ for (int i = 0; i < totalGroups.size(); i++) { totalGroups[i] -= counts[it->second][i]; counts[it->second][i] = 0; }
+ }
+ int thisTotal = totals[it->second]; totals[it->second] = 0;
+ total -= thisTotal;
+ indexNameMap.erase(it);
+ }else {
+ m->mothurOut("[ERROR]: Your count table contains does not include " + seqName + ", cannot remove."); m->mothurOutEndLine(); m->control_pressed = true;
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "push_back");
+ exit(1);
+ }
+}
+/************************************************************/
//add seqeunce without group info
int CountTable::push_back(string seqName, int thisTotal) {
try {
if ((hasGroups) && (groupCounts.size() != getNumGroups())) { m->mothurOut("[ERROR]: Your count table has a " + toString(getNumGroups()) + " groups and " + seqName + " has " + toString(groupCounts.size()) + ", please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
int thisTotal = 0;
for (int i = 0; i < getNumGroups(); i++) { totalGroups[i] += groupCounts[i]; thisTotal += groupCounts[i]; }
+ if (hasGroups) { counts.push_back(groupCounts); }
indexNameMap[seqName] = uniques;
totals.push_back(thisTotal);
total+= thisTotal;
}
}
/************************************************************/
-//returns names of seqs
+//returns the names of all unique sequences in file
+vector<string> CountTable::getNamesOfSeqs(string group) {
+ try {
+ vector<string> names;
+ if (hasGroups) {
+ map<string, int>::iterator it = indexGroupMap.find(group);
+ if (it == indexGroupMap.end()) {
+ m->mothurOut("[ERROR]: " + group + " is not in your count table. Please correct.\n"); m->control_pressed = true;
+ }else {
+ for (map<string, int>::iterator it2 = indexNameMap.begin(); it2 != indexNameMap.end(); it2++) {
+ if (counts[it2->second][it->second] != 0) { names.push_back(it2->first); }
+ }
+ }
+ }else{ m->mothurOut("[ERROR]: Your count table does not have group info. Please correct.\n"); m->control_pressed = true; }
+
+ return names;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "getNamesOfSeqs");
+ exit(1);
+ }
+}
+/************************************************************/
+//merges counts of seq1 and seq2, saving in seq1
int CountTable::mergeCounts(string seq1, string seq2) {
try {
map<string, int>::iterator it = indexNameMap.find(seq1);
m->mothurOut("[ERROR]: " + seq2 + " is not in your count table. Please correct.\n"); m->control_pressed = true;
}else {
//merge data
- for (int i = 0; i < groups.size(); i++) {
- counts[it->second][i] += counts[it2->second][i];
- counts[it2->second][i] = 0;
- }
+ for (int i = 0; i < groups.size(); i++) { counts[it->second][i] += counts[it2->second][i]; }
totals[it->second] += totals[it2->second];
- totals[it2->second] = 0;
uniques--;
indexNameMap.erase(it2);
}
}
-
return 0;
}
catch(exception& e) {
exit(1);
}
}
+/************************************************************/
+int CountTable::copy(CountTable* ct) {
+ try {
+ vector<string> thisGroups = ct->getNamesOfGroups();
+ for (int i = 0; i < thisGroups.size(); i++) { addGroup(thisGroups[i]); }
+ vector<string> names = ct->getNamesOfSeqs();
+
+ for (int i = 0; i < names.size(); i++) {
+ vector<int> thisCounts = ct->getGroupCounts(names[i]);
+ push_back(names[i], thisCounts);
+ }
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "CountTable", "copy");
+ exit(1);
+ }
+}
/************************************************************/
projects / mothur.git / blobdiff