Sep 18, 2024: README and cleaning

README.md

@@ -6,15 +6,15 @@ Teloscope is a fast and comprehensive tool for matching, counting and reporting
 * GC
 * Shannon Entropy
 
-Teloscope reports all these metrics and the final telomere coordinates in BED/BEDgraph files and produces a summary report.
+Teloscope reports all these metrics in BED/BEDgraph files and produces a summary report.
 
 ## Installation
 
 Either download one of the releases or `git clone https://github.com/vgl-hub/teloscope.git --recursive` and `make -j` in `teloscope` folder.
 
 ## Usage
 
-`teloscope -f input.[fasta][.gz] -p TTAGGG,TTAGGGG -w [window size] -s [step size]`
+`teloscope -f input.[fasta][.gz] -o [output/dir] -p TTAGGG,TTAGGGG -w [window size] -s [step size] -k`
 
 To check out all options and flags, please use:
 `teloscope -h`

include/input.h

@@ -11,9 +11,8 @@ struct UserInputTeloscope : UserInput {
     uint32_t step = 500;
     double maxMem = 0;
     std::string prefix = ".", outFile = "";
-    bool keepWindowData = false;
+    bool keepWindowData = false; // Memory intensive
     bool modeMatch = true, modeEntropy = true, modeGC = true; // Change to: de novo, user-defined
-    bool storeWindowData = false; // Memory intensive
     float densityThreshold = 0.5f; // Threshold for telomere detection
     uint32_t maxGaps = 2;   // Allowed gaps in telomere blocks
     uint8_t mergeDistance = 6;

include/teloscope.h

@@ -9,7 +9,7 @@
 #include <memory>
 #include <unordered_map>
 
-std::string removeCarriageReturns(const std::string& input);
+// std::string removeCarriageReturns(const std::string& input);
 
 class Trie {
     struct TrieNode {
@@ -61,6 +61,7 @@ struct PatternData {
     float density = 0.0f; // Density of the pattern
 };
 
+
 struct WindowData {
     uint32_t windowStart;
     uint32_t currentWindowSize;
@@ -124,7 +125,7 @@ class Teloscope {
                     std::unordered_map<std::string, std::ofstream>& patternCountFiles,
                     std::unordered_map<std::string, std::ofstream>& patternDensityFiles) {
 
-        if (!userInput.storeWindowData) { // If windowData is not stored, return
+        if (!userInput.keepWindowData) { // If windowData is not stored, return
             return;
         }
 
@@ -202,8 +203,8 @@ class Teloscope {
         }
 
         if (userInput.modeMatch) {
-            telomereBEDFile.open(outRoute + "/telomere_blocks.bed"); // CHECK
-            telomereCountFile.open(outRoute + "/telomere_block_counts.txt"); // CHECK
+            // telomereBEDFile.open(outRoute + "/telomere_blocks.bed"); // CHECK
+            // telomereCountFile.open(outRoute + "/telomere_block_counts.txt"); // CHECK
 
             for (const auto& pattern : userInput.patterns) {
                 patternMatchFiles[pattern].open(outRoute + "/" + pattern + "_matches.bed");
@@ -225,8 +226,8 @@ class Teloscope {
             gcContentFile.close();
         }
         if (userInput.modeMatch) {
-            telomereBEDFile.close(); // CHECK
-            telomereCountFile.close(); // CHECK
+            // telomereBEDFile.close(); // CHECK
+            // telomereCountFile.close(); // CHECK
 
             for (auto& [pattern, file] : patternMatchFiles) {
                 file.close();
@@ -251,7 +252,7 @@ class Teloscope {
 
         // For each pattern, print the path header with the highest number of matches - PENDING
         // For each pattern, print the path header with the lowest number of matches - PENDING
-        if (userInput.storeWindowData) {
+        if (userInput.keepWindowData) {
             std::cout << "Max Shannon Entropy:\t" << getMax(entropyValues) << "\n";
             std::cout << "Mean Shannon Entropy:\t" << getMean(entropyValues) << "\n";
             std::cout << "Median Shannon Entropy:\t" << getMedian(entropyValues) << "\n";

src/input.cpp

@@ -69,15 +69,16 @@ bool Teloscope::walkPath(InPath* path, std::vector<InSegment*> &inSegments, std:
     unsigned int cUId = 0, gapLen = 0, seqPos = path->getSeqPos();
     std::vector<PathComponent> pathComponents = path->getComponents();
 
-    std::string header = removeCarriageReturns(path->getHeader());
+    // std::string header = removeCarriageReturns(path->getHeader());
+    std::string header = path->getHeader();
+    std::cout << "header_before: " << header << std::endl;
+    eraseChar(header, '\r');
+    std::cout << "header_after: " << header << std::endl;
     threadLog.add("\n\tWalking path:\t" + path->getHeader());
 
     std::vector<WindowData> pathWindows;
     std::vector<TelomereBlock> pathTelomereBlocks;
 
-    // std::string header = path->getHeader();
-    // eraseChar(header, '\r');
-
     for (std::vector<PathComponent>::iterator component = pathComponents.begin(); component != pathComponents.end(); component++) {
 
         cUId = component->id;

src/main.cpp

@@ -61,7 +61,7 @@ int main(int argc, char **argv) {
 
         int option_index = 0;
 
-        c = getopt_long(argc, argv, "-:f:j:m:o:p:s:w:vh", long_options, &option_index);
+        c = getopt_long(argc, argv, "-:f:j:m:o:p:s:w:kvh", long_options, &option_index);
 
         // if (optind < argc && !isPipe) { // if pipe wasn't assigned already
 

src/teloscope.cpp

@@ -12,21 +12,16 @@
 #include "log.h"
 #include "global.h"
 #include "uid-generator.h"
-
 #include "bed.h"
 #include "struct.h"
 #include "functions.h"
-
 #include "gfa-lines.h"
 #include "gfa.h"
 #include "sak.h"
-
 #include "stream-obj.h"
-
 #include "input-agp.h"
 #include "input-filters.h"
 #include "input-gfa.h"
-
 #include "teloscope.h"
 #include "input.h"
 
@@ -46,17 +41,6 @@ void Trie::insertPattern(const std::string& pattern) {
 }
 
 
-std::string removeCarriageReturns(const std::string& input) {
-    std::string output = input;
-    output.erase(std::remove(output.begin(), output.end(), '\r'), output.end());
-    return output;
-}
-
-// void eraseChar(std::string& input, char rmChar) {
-//     input.erase(std::remove(input.begin(), input.end(), rmChar), input.end());
-// }
-
-
 float Teloscope::getShannonEntropy(const std::unordered_map<char, uint32_t>& nucleotideCounts, uint32_t windowSize) {
     float entropy = 0.0;
     for (auto &[nucleotide, count] : nucleotideCounts) {
@@ -107,7 +91,7 @@ float Teloscope::getMax(const std::vector<float> values) {
 
 
 void Teloscope::insertWindowData(unsigned int seqPos, const std::string& header, std::vector<WindowData>& pathWindows) {
-    if (userInput.storeWindowData) {
+    if (userInput.keepWindowData) {
         allWindows.push_back(std::make_tuple(seqPos, header, pathWindows));
     }
 }