Fix truncated profile sequences in convertalis #567

soedinglab · Jun 2, 2022 · b0b8e85 · b0b8e85
1 parent 96b2009
commit b0b8e85
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 11 deletions.
diff --git a/src/commons/Sequence.cpp b/src/commons/Sequence.cpp
@@ -375,18 +375,18 @@ void Sequence::print() {
     std::cout << std::endl;
 }
 
-void extractProfileData(const char* data, const BaseMatrix &subMat, const int offset, std::string &result) {
+void extractProfileData(const char* data, size_t dataSize, const BaseMatrix &subMat, const int offset, std::string &result) {
     size_t i = 0;
-    while (data[i] != '\0'){
+    while (i < dataSize){
         result.append(1, subMat.num2aa[(int)data[i + Sequence::PROFILE_AA_SIZE + offset]]);
         i += Sequence::PROFILE_READIN_SIZE;
     }
 }
 
-void Sequence::extractProfileSequence(const char* data, const BaseMatrix &submat, std::string &result) {
-    extractProfileData(data, submat, 0, result);
+void Sequence::extractProfileSequence(const char* data, size_t dataSize, const BaseMatrix &submat, std::string &result) {
+    extractProfileData(data, dataSize, submat, 0, result);
 }
 
-void Sequence::extractProfileConsensus(const char* data, const BaseMatrix &submat, std::string &result) {
-    extractProfileData(data, submat, 1, result);
+void Sequence::extractProfileConsensus(const char* data, size_t dataSize, const BaseMatrix &submat, std::string &result) {
+    extractProfileData(data, dataSize, submat, 1, result);
 }
diff --git a/src/commons/Sequence.h b/src/commons/Sequence.h
@@ -417,8 +417,8 @@ class Sequence {
 
     void print(); // for debugging
 
-    static void extractProfileSequence(const char* data, const BaseMatrix &submat, std::string &result);
-    static void extractProfileConsensus(const char* data, const BaseMatrix &submat, std::string &result);
+    static void extractProfileSequence(const char* data, size_t dataSize, const BaseMatrix &submat, std::string &result);
+    static void extractProfileConsensus(const char* data, size_t dataSize, const BaseMatrix &submat, std::string &result);
 
     int getId() const { return id; }
 

diff --git a/src/util/convertalignments.cpp b/src/util/convertalignments.cpp
@@ -366,7 +366,8 @@ int convertalignments(int argc, const char **argv, const Command &command) {
                     querySeqData = (char*) queryBuffer.c_str();
                 }
                 if (queryProfile) {
-                    Sequence::extractProfileConsensus(querySeqData, *subMat, queryProfData);
+                    size_t queryEntryLen = qDbr.sequenceReader->getEntryLen(qId);
+                    Sequence::extractProfileConsensus(querySeqData, queryEntryLen, *subMat, queryProfData);
                 }
             }
 
@@ -481,7 +482,8 @@ int convertalignments(int argc, const char **argv, const Command &command) {
                                 size_t tId = tDbr->sequenceReader->getId(res.dbKey);
                                 targetSeqData = tDbr->sequenceReader->getData(tId, thread_idx);
                                 if (targetProfile) {
-                                    Sequence::extractProfileConsensus(targetSeqData, *subMat, targetProfData);
+                                    size_t targetEntryLen = tDbr->sequenceReader->getEntryLen(tId);
+                                    Sequence::extractProfileConsensus(targetSeqData, targetEntryLen, *subMat, targetProfData);
                                 }
                             }
                             for(size_t i = 0; i < outcodes.size(); i++) {
@@ -723,7 +725,8 @@ int convertalignments(int argc, const char **argv, const Command &command) {
                         size_t tId = tDbr->sequenceReader->getId(res.dbKey);
                         char* targetSeqData = tDbr->sequenceReader->getData(tId, thread_idx);
                         if (targetProfile) {
-                            Sequence::extractProfileConsensus(targetSeqData, *subMat, targetProfData);
+                            size_t targetEntryLen = tDbr->sequenceReader->getEntryLen(tId);
+                            Sequence::extractProfileConsensus(targetSeqData, targetEntryLen, *subMat, targetProfData);
                             printSeqBasedOnAln(result, targetProfData.c_str(), res.dbStartPos,
                                                Matcher::uncompressAlignment(res.backtrace), true,
                                                (res.dbStartPos > res.dbEndPos),