Merge pull request #14 from vdanjean/master

Fix a deadlock with multi threads and various improvement proposed

Author: sfchen

.gitignore

@@ -10,6 +10,9 @@ mutscan
 *.o
 *.obj
 
+# Dependency files
+*.d
+
 # Precompiled Headers
 *.gch
 *.pch

Makefile

@@ -11,24 +11,24 @@ TARGET = mutscan
 BIN_TARGET = ${TARGET}
 
 CC = g++
+CPPFLAGS = -Wall
 CFLAGS = -std=c++11 -g -I${DIR_INC}
 
 ${BIN_TARGET}:${OBJ}
 	$(CC) $(OBJ) -lz -lpthread -o $@
 
-${DIR_OBJ}/%.o:${DIR_SRC}/%.cpp make_obj_dir
-	$(CC) $(CFLAGS) -O3 -c  $< -o $@
+${DIR_OBJ}/%.o:${DIR_SRC}/%.cpp
+	@mkdir -p "${DIR_OBJ}"
+	$(CC) $(CPPFLAGS) $(CFLAGS) -MMD -O3 -c  $< -o $@
+
+-include $(wildcard ${DIR_OBJ}/*.d)
+
 .PHONY:clean
 clean:
 	rm obj/*.o
+	rm obj/*.d
 	rm mutscan
 
-make_obj_dir:
-	@if test ! -d $(DIR_OBJ) ; \
-	then \
-		mkdir $(DIR_OBJ) ; \
-	fi
-
 install:
-	install $(TARGET) $(BINDIR)/$(TARGET)
-	@echo "Installed."
+	install $(TARGET) $(DESTDIR)$(BINDIR)/$(TARGET)
+	@echo "Installed."

src/editdistance.cpp

@@ -42,7 +42,7 @@ unsigned int edit_distance_bpv(T &cmap, char const *vec, size_t const &vecsize,
     for(size_t i = 0; i < tlen; ++i) VP[tmax] |= (1L << i);
     for(size_t i = 0; i < vecsize; ++i) {
         TVALUE &PM = cmap[vec[i]];
-        for(int r = 0; r <= tmax; ++r) {
+        for(unsigned int r = 0; r <= tmax; ++r) {
             uint64_t X = PM[r];
             if(r > 0 && (HN[r - 1] & lmb)) X |= 1L;
             D0[r] = (((X & VP[r]) + VP[r]) ^ VP[r]) | X | VN[r];
@@ -65,10 +65,10 @@ unsigned int edit_distance_bpv(T &cmap, char const *vec, size_t const &vecsize,
 template<typename T>
 unsigned int edit_distance_dp(T const *str1, size_t const size1, T const *str2, size_t const size2) {
     vector< vector<uint32_t> > d(size1 + 1, vector<uint32_t>(size2 + 1));
-    for (int i = 0; i < size1 + 1; i++) d[i][0] = i;
-    for (int i = 0; i < size2 + 1; i++) d[0][i] = i;
-    for (int i = 1; i < size1 + 1; i++) {
-        for (int j = 1; j < size2 + 1; j++) {
+    for (size_t i = 0; i < size1 + 1; i++) d[i][0] = i;
+    for (size_t i = 0; i < size2 + 1; i++) d[0][i] = i;
+    for (size_t i = 1; i < size1 + 1; i++) {
+        for (size_t j = 1; j < size2 + 1; j++) {
             d[i][j] = min(min(d[i-1][j], d[i][j-1]) + 1, d[i-1][j-1] + (str1[i-1] == str2[j-1] ? 0 : 1));
         }
     }
@@ -131,7 +131,7 @@ unsigned int edit_distance(string a, string b) {
 
 unsigned int hamming_distance(const char *a, const unsigned int asize, const char *b, const unsigned int bsize) {
     int dis = 0;
-    for(int i=0; i<min(asize, bsize); i++) {
+    for(unsigned int i=0; i<min(asize, bsize); i++) {
         if(a[i]!=b[i])
             dis++;
     }

src/htmlreporter.cpp

@@ -43,9 +43,9 @@ void HtmlReporter::printHelper() {
 void HtmlReporter::printMutations() {
     // calculate the found mutation
     int found = 0;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         vector<Match*> matches = mMutationMatches[i];
-        if(matches.size()>=GlobalSettings::minReadSupport){
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport){
             found++;
         }
     }
@@ -56,9 +56,9 @@ void HtmlReporter::printMutations() {
         if(found>1)
             mFile<<"s";
         mFile<<":</p><ul>";
-        for(int i=0;i<mMutationList.size();i++){
+        for(size_t i=0;i<mMutationList.size();i++){
             vector<Match*> matches = mMutationMatches[i];
-            if(matches.size()>=GlobalSettings::minReadSupport){
+            if((ssize_t)matches.size()>=GlobalSettings::minReadSupport){
                 id++;
                 mFile<<"<li class='menu_item'><a href='#"<<mMutationList[i].mName<<"'> " << id << ", " << mMutationList[i].mName;
                 mFile<< " (" << matches.size() << " reads support, " << Match::countUnique(matches) << " unique)" << "</a></li>";
@@ -67,9 +67,9 @@ void HtmlReporter::printMutations() {
         mFile<<"</ul></div>";
     }
     id=0;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         vector<Match*> matches = mMutationMatches[i];
-        if(matches.size()>=GlobalSettings::minReadSupport){
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport){
             id++;
             printMutation(id, mMutationList[i], matches);
         }
@@ -80,11 +80,11 @@ void HtmlReporter::printMutationsJS() {
     mFile << "\n<script type=\"text/javascript\">" << endl;
     mFile << "var data_break = [";
     int id=0;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         vector<Match*> matches = mMutationMatches[i];
-        if(matches.size()>=GlobalSettings::minReadSupport){
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport){
             mFile << "\n[";
-            for(int m=0; m<matches.size(); m++){
+            for(size_t m=0; m<matches.size(); m++){
                 mFile << "\n[";
                 matches[m]->printJS(mFile, mMutationList[i].mLeft.length(), mMutationList[i].mCenter.length(), mMutationList[i].mRight.length());
                 mFile << "],"; 
@@ -140,7 +140,7 @@ void HtmlReporter::printMutation(int id, Mutation& mutation, vector<Match*>& mat
     mFile << "<td>" << mutation.mRight << "</td>";
     mFile << "<td>" << "" << "</td>";
     mFile << "</tr>";
-    for(int m=0; m<matches.size(); m++){
+    for(size_t m=0; m<matches.size(); m++){
         long rowid = id*100000 + m;
         if(!GlobalSettings::simplifiedMode)
             mFile << "<tr onclick='toggle(" << rowid << ");'>";
@@ -266,7 +266,7 @@ void HtmlReporter::printScanTargets(){
     mFile << "<p> scanned " << mMutationList.size() << " mutation spots...<input type='button' id='target_view_btn', onclick=toggle_target_list('target_list'); value='show'></input></p>";
     mFile << "<ul id='target_list' style='display:none'>";
     int id=0;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         id++;
         mFile<<"<li> " << id << ", " << mMutationList[i].mName << "</li>";
     }

src/jsonreporter.cpp

@@ -25,10 +25,10 @@ void JsonReporter::run() {
     mFile << "\t\"mutations\":{";
 
     bool isFirstMut = true;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         Mutation mut = mMutationList[i];
         vector<Match*> matches = mMutationMatches[i];
-        if(matches.size()>=GlobalSettings::minReadSupport){
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport){
             if(isFirstMut) {
                 mFile << endl;
                 isFirstMut = false;
@@ -40,7 +40,7 @@ void JsonReporter::run() {
             mFile << "\t\t\t\"" <<  "chr" << "\":" << "\"" << mut.mChr << "\"," << endl;
             mFile << "\t\t\t\"" <<  "ref" << "\":[" << "\"" << mut.mLeft << "\"," << "\"" << mut.mCenter << "\"," << "\"" << mut.mRight << "\"]," << endl;
             mFile << "\t\t\t\"" <<  "reads" << "\":[" << endl;
-            for(int m=0; m<matches.size(); m++){
+            for(size_t m=0; m<matches.size(); m++){
                 mFile << "\t\t\t\t{" << endl;
                 mFile << "\t\t\t\t\t\"breaks\":"; 
                 matches[m]->printBreaksToJson(mFile, mut.mLeft.length(), mut.mCenter.length(), mut.mRight.length());

src/match.cpp

@@ -34,7 +34,7 @@ Match::~Match(){
     // we don't delete mRead or mSequence here since they are shared by different objects
     // and will be deleted in other places
     if(mOriginalReads) {
-        for(int i=0;i<mOriginalReads->size();i++){
+        for(size_t i=0;i<mOriginalReads->size();i++){
             delete (*mOriginalReads)[i];
             (*mOriginalReads)[i] = NULL;
         }
@@ -112,7 +112,7 @@ void Match::printBreaksToJson(ofstream& file, int leftlen, int centerlen, int ri
     breaks.push_back( mPos+centerlen );
     breaks.push_back( min(mPos+centerlen+rightlen, mReadLen));
     file << "[";
-    for(int i=0; i<breaks.size(); i++) {
+    for(size_t i=0; i<breaks.size(); i++) {
         file << breaks[i];
         if(i!=breaks.size() - 1) 
             file << ",";
@@ -149,7 +149,7 @@ void Match::printJS(ofstream& file, int leftlen, int centerlen, int rightlen) {
 void Match::printReadsToFile(ofstream& file){
     if(!mOriginalReads)
         return;
-    for(int i=0;i<mOriginalReads->size();i++){
+    for(size_t i=0;i<mOriginalReads->size();i++){
         (*mOriginalReads)[i]->printFile(file);
     }
 }
@@ -163,7 +163,7 @@ int Match::countUnique(vector<Match*>& matches) {
         return 0;
     int count = 1;
     Match* cur = matches[0];
-    for(int i=1;i<matches.size();i++){
+    for(size_t i=1;i<matches.size();i++){
         Match* m = matches[i];
         if( *m > *cur || *m < *cur) {
             cur = m;

src/match.h

@@ -27,6 +27,7 @@ class Match{
     void setReversed(bool flag);
     void addOriginalRead(Read* r);
     void addOriginalPair(ReadPair* pair);
+    template <class ReadData> void addOriginalReadData(ReadData* r);
     int readlength() const;
 
     inline bool operator <(const Match& other) const 
@@ -62,5 +63,14 @@ class Match{
     char mMeanQual;
 };
 
+template <>
+inline void Match::addOriginalReadData<Read> (Read* r) {
+    addOriginalRead(r);
+}
 
-#endif
+template <>
+inline void Match::addOriginalReadData<ReadPair> (ReadPair* pair) {
+    addOriginalPair(pair);
+}
+
+#endif

src/multihtmlreporter.cpp

@@ -20,9 +20,9 @@ MultiHtmlReporter::~MultiHtmlReporter(){
 
 void MultiHtmlReporter::stat(){
     mTotalCount = 0;
-    for(int m=0; m<mMutationList.size(); m++) {
+    for(size_t m=0; m<mMutationList.size(); m++) {
         vector<Match*> matches = mMutationMatches[m];
-        if(matches.size()>=GlobalSettings::minReadSupport) {
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport) {
             mTotalCount++;
             string chr = mMutationList[m].mChr;
             if(mChrCount.count(chr)==0)
@@ -69,9 +69,9 @@ void MultiHtmlReporter::printAllChromosomeLink(ofstream& file) {
     map<string, int>::iterator iter;
     file << "<ul id='menu'>";
     file << "<div style='font-size:12px;padding-top:20px;text-align:left;color:#aaaaaa'>Mutations found of all chromosomes:</div>";
-    for(int m=0; m<mMutationList.size(); m++) {
+    for(size_t m=0; m<mMutationList.size(); m++) {
         vector<Match*> matches = mMutationMatches[m];
-        if(matches.size()>=GlobalSettings::minReadSupport) {
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport) {
             found = true;
             string chr = mMutationList[m].mChr;
             string filename = chr + "/" + to_string(m) + ".html";
@@ -88,9 +88,9 @@ void MultiHtmlReporter::printAllChromosomeLink(ofstream& file) {
 
 void MultiHtmlReporter::printChrLink(ofstream& file, string chr) {
     bool found = false;
-    for(int m=0; m<mMutationList.size(); m++) {
+    for(size_t m=0; m<mMutationList.size(); m++) {
         vector<Match*> matches = mMutationMatches[m];
-        if(matches.size()>=GlobalSettings::minReadSupport) {
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport) {
             found = true;
             if(chr == mMutationList[m].mChr) {
                 string filename = chr + "/" + to_string(m) + ".html";
@@ -106,9 +106,9 @@ void MultiHtmlReporter::printChrLink(ofstream& file, string chr) {
 }
 
 void MultiHtmlReporter::printMutationHtml() {
-    for(int m=0; m<mMutationList.size(); m++) {
+    for(size_t m=0; m<mMutationList.size(); m++) {
         vector<Match*> matches = mMutationMatches[m];
-        if(matches.size()>=GlobalSettings::minReadSupport) {
+        if((ssize_t)matches.size()>=GlobalSettings::minReadSupport) {
             string chr = mMutationList[m].mChr;
             string folder = mFolderPath + "/" + chr;
             string filename = folder + "/" + to_string(m) + ".html";
@@ -262,7 +262,7 @@ void MultiHtmlReporter::printScanTargets(ofstream& file){
     file << "<p> scanned " << mMutationList.size() << " mutation spots...<input type='button' id='target_view_btn', onclick=toggle_target_list('target_list'); value='show'></input></p>";
     file << "<ul id='target_list' style='display:none'>";
     int id=0;
-    for(int i=0;i<mMutationList.size();i++){
+    for(size_t i=0;i<mMutationList.size();i++){
         id++;
         file<<"<li> " << id << ", " << mMutationList[i].mName << "</li>";
     }

src/mutation.cpp

@@ -181,7 +181,7 @@ vector<Mutation> Mutation::parseBuiltIn() {
     vector<Mutation> mutations;
     vector<string> lines;
     split(BUILT_IN_MUTATIONS, lines, "\n");
-    for(int i=0;i<lines.size();i++){
+    for(size_t i=0;i<lines.size();i++){
         string linestr = lines[i];
         vector<string> splitted;
         split(linestr, splitted, ",");
@@ -228,7 +228,7 @@ vector<Mutation> Mutation::parseVcf(string vcfFile, string refFile) {
     fr.readAll();
     map<string, string> ref = fr.contigs();
 
-    for(int i=0;i<variants.size();i++) {
+    for(size_t i=0;i<variants.size();i++) {
         Variant& v = variants[i];
         // skip the unmasked if markedOnly flag is set true
         if(markedOnly && (v.filter!="m" && v.filter!="M"))
@@ -248,7 +248,7 @@ vector<Mutation> Mutation::parseVcf(string vcfFile, string refFile) {
         }
         // the variant is out of this contig, or in the front or tail
         // note that VCF is 1-based, and string is 0-based
-        if(v.pos > ref[chrom].length() + 25 + v.ref.length() || v.pos < 25 + v.ref.length())
+        if((size_t)v.pos > ref[chrom].length() + 25 + v.ref.length() || (size_t)v.pos < 25 + v.ref.length())
             continue;
 
         string gene = v.gene();