Nov 22, 2024: Simplify functions

vgl-hub · Nov 22, 2024 · 970b97f · 970b97f
1 parent c347b76
commit 970b97f
Show file tree

Hide file tree

Showing 4 changed files with 116 additions and 191 deletions.
diff --git a/include/input.h b/include/input.h
@@ -13,8 +13,9 @@ struct UserInputTeloscope : UserInput {
     uint32_t windowSize = 1000;
     uint8_t kmerLen = 21;
     uint32_t step = 500;
-    unsigned short int minBlockLen = 50;
-    unsigned short int maxBlockDist = 50;
+    unsigned short int minBlockLen = 100; // Not used anymore
+    unsigned short int maxBlockDist = 200;
+    unsigned short int minBlockCounts = 2;
 
     bool keepWindowData = false; // Memory intensive
     bool modeMatch = true, modeEntropy = true, modeGC = true; // Change to: de novo, user-defined

diff --git a/include/teloscope.h b/include/teloscope.h
@@ -49,6 +49,8 @@ class Trie {
 struct TelomereBlock {
     uint64_t start;
     uint16_t blockLen; // End = start + blockLen
+    uint32_t blockDistance;
+    uint16_t blockCounts;
 };
 
 struct WindowData {
@@ -60,11 +62,6 @@ struct WindowData {
     // uint32_t winHDistance = 0;
 
     std::vector<uint8_t> hDistances; 
-    // std::vector<TelomereBlock> winBlocks;
-
-    // std::vector<uint32_t> canonicalMatches;
-    // std::vector<uint32_t> nonCanonicalMatches;
-    // std::vector<uint32_t> windowMatches;
     uint16_t canonicalCounts = 0;
     uint16_t nonCanonicalCounts = 0;
     float canonicalDensity = 0.0f;
@@ -88,6 +85,8 @@ struct PathData {
     std::string header;
     std::vector<WindowData> windows; // Empty unless specified by user
     std::unordered_map<std::string, std::vector<TelomereBlock>> mergedBlocks;
+    std::vector<uint32_t> canonicalMatches;
+    std::vector<uint32_t> nonCanonicalMatches;
 };
 
 
@@ -134,7 +133,6 @@ class Teloscope {
 
     bool walkPath(InPath* path, std::vector<InSegment*> &inSegments, std::vector<InGap> &inGaps);
 
-    // void analyzeWindow(const std::string &window, uint32_t windowStart, WindowData& windowData, WindowData& nextOverlapData);
     void analyzeWindow(const std::string &window, uint32_t windowStart, WindowData& windowData, WindowData& nextOverlapData, SegmentData& segmentData);
 
     SegmentData analyzeSegment(std::string &sequence, UserInputTeloscope userInput, uint64_t absPos);
@@ -143,14 +141,11 @@ class Teloscope {
 
     void sortBySeqPos();
 
-    // std::vector<TelomereBlock> getTelomereBlocks(const std::vector<uint32_t>& inputMatches, uint64_t windowStart);
-    std::vector<TelomereBlock> getTelomereBlocks(const std::vector<uint32_t>& inputMatches, uint64_t windowStart, uint32_t currentWindowSize);
-
-    std::vector<TelomereBlock> mergeTelomereBlocks(const std::vector<TelomereBlock>& winBlocks);
+    std::vector<TelomereBlock> getTelomereBlocks(const std::vector<uint32_t>& inputMatches, uint16_t mergeDist);
 
     void writeBEDFile(std::ofstream& windowMetricsFile, std::ofstream& windowRepeatsFile,
                                 std::ofstream& canonicalMatchFile, std::ofstream& noncanonicalMatchFile,
-                                std::ofstream& allBlocksFile, std::ofstream& canonicalBlocksFile, std::ofstream& noncanonicalBlocksFile);
+                                std::ofstream& allBlocksFile, std::ofstream& canonicalBlocksFile);
 
     void handleBEDFile();
 

diff --git a/src/input.cpp b/src/input.cpp
@@ -1,4 +1,5 @@
 #include <iostream>
+#include <stdint.h>
 #include <stdlib.h>
 #include <string>
 #include <stdexcept> // jack: std::runtime_error
@@ -64,16 +65,18 @@ bool Teloscope::walkPath(InPath* path, std::vector<InSegment*> &inSegments, std:
     uint64_t absPos = 0;
     unsigned int cUId = 0, gapLen = 0, seqPos = path->getSeqPos();
     std::vector<PathComponent> pathComponents = path->getComponents();
-    // uint64_t pathLen = path->getLen(); // PENDING
+    uint64_t pathSize = path->getLen();
 
     threadLog.add("\n\tWalking path:\t" + path->getHeader());
     std::string header = path->getHeader();
     eraseChar(header, '\r');
+    uint32_t numSegments = (pathComponents.size() + 1) / 2;
 
     // Initialize PathData for this path
     PathData pathData;
     pathData.seqPos = seqPos;
     pathData.header = header;
+    pathData.windows.reserve((pathSize - (userInput.windowSize + 2 * userInput.step) * (numSegments)) / userInput.step);
 
     for (std::vector<PathComponent>::iterator component = pathComponents.begin(); component != pathComponents.end(); component++) {
 
@@ -88,17 +91,35 @@ bool Teloscope::walkPath(InPath* path, std::vector<InSegment*> &inSegments, std:
             if (component->orientation == '+') {
                 SegmentData segmentData = analyzeSegment(sequence, userInput, absPos);
 
-                if (userInput.keepWindowData) {
-                    pathData.windows.insert(pathData.windows.end(), segmentData.windows.begin(), segmentData.windows.end());
-                }
+                // Collect window data
+                pathData.windows.insert(
+                    pathData.windows.end(),
+                    std::make_move_iterator(segmentData.windows.begin()),
+                    std::make_move_iterator(segmentData.windows.end())
+                );
 
-                for (const auto& [groupName, blocks] : segmentData.mergedBlocks) {
+                // Collect blocks
+                for (auto& [groupName, blocks] : segmentData.mergedBlocks) {
                     pathData.mergedBlocks[groupName].insert(
                         pathData.mergedBlocks[groupName].end(),
-                        blocks.begin(),
-                        blocks.end()
+                        std::make_move_iterator(blocks.begin()),
+                        std::make_move_iterator(blocks.end())
                     );
                 }
+
+                // Collect matches
+                pathData.canonicalMatches.insert(
+                    pathData.canonicalMatches.end(),
+                    std::make_move_iterator(segmentData.canonicalMatches.begin()),
+                    std::make_move_iterator(segmentData.canonicalMatches.end())
+                );
+
+                pathData.nonCanonicalMatches.insert(
+                    pathData.nonCanonicalMatches.end(),
+                    std::make_move_iterator(segmentData.nonCanonicalMatches.begin()),
+                    std::make_move_iterator(segmentData.nonCanonicalMatches.end())
+                );
+
             } else {
             }