major changes

Removal of species selection. Now supporting all species with peptide and gtf files available from Ensembl. Adding new parameter -genome that takes a genome fasta file (nucleotide sequences of chromosomes/scaffolds etc) to inform separation of chromosome (primary assembly) and scaffolds in the output files. If not provided the coordinates will be extracted solely from the GTF file.
Bug fix for multiple input files: all peptides now mapped

Author: cschlaffner

PoGo/src/Chromosome.cpp

@@ -0,0 +1,106 @@
+#include "Chromosome.h"
+
+std::map<std::string, int> Chromosome::chr_to_int;
+std::map<int, std::string> Chromosome::int_to_chr;
+std::map<std::string, int> Chromosome::scaffold_names;
+
+Chromosome::Chromosome(void) {
+	this->name = "NA";
+}
+
+Chromosome::Chromosome(std::string const & name) {
+	this->name = name;
+}
+
+Chromosome::~Chromosome(void) {
+
+}
+
+std::map<std::string, int>& Chromosome::getChr_to_int(void) {
+	return chr_to_int;
+}
+
+std::map<int, std::string>& Chromosome::getInt_to_chr(void) {
+	return int_to_chr;
+}
+
+std::map<std::string, int>& Chromosome::getScaffold_names(void) {
+	return scaffold_names;
+}
+
+std::string Chromosome::forValue(int const & value) {
+	std::map<int, std::string>::iterator it = getInt_to_chr().find(value);
+	if (it != getInt_to_chr().end()) {
+		return it->second;
+	}
+	return "NA";
+}
+
+int Chromosome::forName(std::string const & name) {
+	std::map<std::string, int>::iterator it = getChr_to_int().find(name);
+	if (it != getChr_to_int().end()) {
+		return it->second;
+	}
+	return -1;
+}
+
+void Chromosome::addChr(std::string const & name) {
+	std::string tmpname = name;
+	std::string substring = tmpname.substr(0, 3);
+	if (substring == "chr" || substring == "Chr") {
+		tmpname = tmpname.substr(3);
+	}
+	std::map<std::string, int>::iterator it = getChr_to_int().find(tmpname);
+	if (it == getChr_to_int().end())
+	{
+		it = getScaffold_names().find(tmpname);
+		if (it == getScaffold_names().end()) {
+			int newrank = getChr_to_int().size() + 1;
+			getChr_to_int().insert(std::pair<std::string, int>(tmpname, newrank));
+			getInt_to_chr().insert(std::pair<int, std::string>(newrank, tmpname));
+			if (tmpname == "M") {
+				newrank = newrank + 1;
+				getChr_to_int().insert(std::pair<std::string, int>("MT", newrank));
+				getInt_to_chr().insert(std::pair<int, std::string>(newrank, "MT"));
+			}
+			else if (tmpname == "MT") {
+				newrank = newrank + 1;
+				getChr_to_int().insert(std::pair<std::string, int>("M", newrank));
+				getInt_to_chr().insert(std::pair<int, std::string>(newrank, "M"));
+			}
+		}
+	}
+}
+
+void Chromosome::addScaffold(std::string const & name) {
+	std::map<std::string, int>::iterator it = getChr_to_int().find("scaffold");
+	if (it != getChr_to_int().end()) {
+		int newrank = getChr_to_int().size() + 1;
+		getChr_to_int().insert(std::pair<std::string, int>("scaffold", newrank));
+	}
+}
+
+int Chromosome::getValue(void) const {
+	std::map<std::string, int>::iterator it = getChr_to_int().find(name);
+	if (it != getChr_to_int().end()) {
+		return it->second;
+	}
+	return -1;
+}
+
+std::string Chromosome::getName(void) const {
+	return name;
+}
+
+void Chromosome::setName(std::string const& name) {
+	this->name = name;
+}
+
+bool Chromosome::isNA(void) const {
+	return name == "NA";
+}
+
+bool Chromosome::isScaffold(void) const {
+	std::map<std::string, int>::iterator it = getScaffold_names().find(name);
+	return it != getScaffold_names().end();
+}

PoGo/src/Chromosome.h

@@ -0,0 +1,40 @@
+#ifndef CHROMOSOME_H
+#define CHROMOSOME_H
+
+#include "Globals.h"
+
+class Chromosome {
+public:
+	//ctr / dtr
+	Chromosome(void);
+	Chromosome(std::string const& name);
+	~Chromosome(void);
+	//end ctr / dtr
+
+	static std::string forValue(int const& value);
+	static int forName(std::string const& name);
+	static void addChr(std::string const& name);
+	static void addScaffold(std::string const& name);
+
+	int getValue(void) const;
+	std::string getName(void) const;
+	void setName(std::string const& name);
+	bool isNA(void) const;
+	bool isScaffold(void) const;
+
+private:
+	//holds mapping of chromosome names to order values
+	static std::map<std::string, int> chr_to_int;
+	//holds mapping of order values to chromosome names
+	static std::map<int, std::string> int_to_chr;
+	//holds scaffold names
+	static std::map<std::string, int> scaffold_names;
+
+	std::string name;
+
+	static std::map<std::string, int>& getChr_to_int(void);
+	static std::map<int, std::string>& getInt_to_chr(void);
+	static std::map<std::string, int>& getScaffold_names(void);
+};
+
+#endif

PoGo/src/CoordinateWrapper.cpp

@@ -11,6 +11,12 @@ int CoordinateWrapper::size() const {
 	return m_map.size();
 }
 
+void CoordinateWrapper::renew() {
+	m_existing_peptides->clear();
+	delete(m_existing_peptides);
+	m_existing_peptides = new ExistingPeptides();
+}
+
 ProteinEntry& CoordinateWrapper::lookup_entry(std::string transcriptId) {
 	return m_map[transcriptId];
 }

PoGo/src/CoordinateWrapper.h

@@ -22,6 +22,8 @@ class CoordinateWrapper {
 	void add(ProteinEntry entry);
 	//returns the number of ProteinEntry elements currently in the CorrdinateWrapper.
 	int size() const;
+
+	void renew();
 
 	//reads and parses a fasta file and adds all of them to the CoordinateWrapper.
 	void read_fasta_file(std::string file);

PoGo/src/Coordinates.h

@@ -1,7 +1,7 @@
 #ifndef COORDINATES_H
 #define COORDINATES_H
 
-#include "Globals.h"
+#include "Chromosome.h"
 
 //possible offsets.
 enum Offset {
@@ -10,107 +10,6 @@ enum Offset {
 	off3 = 3
 };
 
-//possible chromosomes
-enum Chromosome {
-	chr1 = 1,
-	chr1A = 2,
-	chr1B = 3,
-	chr2 = 4,
-	chr2A = 5,
-	chr2a = 6,
-	chr2B = 7,
-	chr2b = 8,
-	chr3 = 9,
-	chr4 = 10,
-	chr4A = 11,
-	chr5 = 12,
-	chr6 = 13,
-	chr7 = 14,
-	chr8 = 15,
-	chr9 = 16,
-	chr10 = 17,
-	chr11 = 18,
-	chr12 = 19,
-	chr13 = 20,
-	chr14 = 21,
-	chr15 = 22,
-	chr16 = 23,
-	chr17 = 24,
-	chr18 = 25,
-	chr19 = 26,
-	chr20 = 27,
-	chr21 = 28,
-	chr22 = 29,
-	chr23 = 30,
-	chr24 = 31,
-	chr25 = 32,
-	chr26 = 33,
-	chr27 = 34,
-	chr28 = 35,
-	chr29 = 36,
-	chr30 = 37,
-	chr31 = 38,
-	chr32 = 39,
-	chr33 = 40,
-	chr34 = 41,
-	chr35 = 42,
-	chr36 = 43,
-	chr37 = 44,
-	chr38 = 45,
-	chr39 = 46,
-	chr40 = 47,
-	chrI = 48,
-	chrII = 49,
-	chrIII = 50,
-	chrIV  = 51,
-	chrV = 52,
-	chrVI = 53,
-	chrVII = 54,
-	chrVIII = 55,
-	chrIX = 56,
-	chrX = 57,
-	chrXI = 58,
-	chrXII = 59,
-	chrXIII = 60,
-	chrXIV = 61,
-	chrXV = 62,
-	chrXVI = 63,
-	chrY = 64,
-	chrXY = 65,
-	chrX1 = 66,
-	chrX2 = 67,
-	chrX3 = 68,
-	chrX5 = 69,
-	chrA1 = 70,
-	chrA2 = 71,
-	chrA3 = 72,
-	chrB1 = 73,
-	chrB2 = 74,
-	chrB3 = 75,
-	chrB4 = 76,
-	chrC1 = 77,
-	chrC2 = 78,
-	chrD1 = 79,
-	chrD2 = 80,
-	chrD3 = 81,
-	chrD4 = 82,
-	chrE1 = 83,
-	chrE2 = 84,
-	chrE3 = 85,
-	chrF1 = 86,
-	chrF2 = 87,
-	chrLG2 = 88,
-	chrLG5 = 89,
-	chrLGE22 = 90,
-	chrW = 91,
-	chrZ = 92,
-	chrM = 93,
-	chrMito = 94,
-	chrLGE64 = 95,
-	chrNA = -1,
-	scaffold = 0
-};
-
 //possible strands.
 enum Strand {
 	fwd = 1,
@@ -137,6 +36,13 @@ struct Coordinates {
 	//cterm offset (see enum Offset)
 	Offset Cterm;
 
+	Coordinates() {
+		start = 0;
+		end = 0;
+		Nterm = Offset::off3;
+		Cterm = Offset::off3;
+	}
+
 	// lt opeator. returns true if lhs.start is lesser than rhs.start and lhs.end is smaller than rhs.end and lhs.end is smaller than rhs.start
 	//otherwise returns false.
 	bool operator()(const Coordinates& lhs, const Coordinates& rhs) const {
@@ -164,52 +70,59 @@ struct GenomeCoordinates : Coordinates {
 	//holds the frame.
 	Frame frame;
 
+	GenomeCoordinates() : Coordinates() {
+		chr = Chromosome();
+		chrscaf = "";
+		strand = Strand::unk;
+		frame = Frame::unknown;
+	}
+
 	bool operator()(const GenomeCoordinates& lhs, const GenomeCoordinates& rhs) const {
-		if (lhs.chr == scaffold && rhs.chr == scaffold && lhs.chrscaf == rhs.chrscaf) {
+		if (lhs.chr.isScaffold() && rhs.chr.isScaffold() && lhs.chrscaf == rhs.chrscaf) {
 			return lhs.start < rhs.start && lhs.end < rhs.end && lhs.end >= rhs.start;
 		}
-		if (lhs.chr == scaffold && rhs.chr == scaffold && lhs.chrscaf != rhs.chrscaf) {
+		if (lhs.chr.isScaffold() && rhs.chr.isScaffold() && lhs.chrscaf != rhs.chrscaf) {
 			return lhs.chrscaf < rhs.chrscaf;
 		}
-		if (lhs.chr == scaffold && rhs.chr != scaffold) {
+		if (lhs.chr.isScaffold() && !rhs.chr.isScaffold()) {
 			return false;
 		}
-		if (lhs.chr != scaffold && rhs.chr == scaffold) {
+		if (!lhs.chr.isScaffold() && rhs.chr.isScaffold()) {
 			return true;
 		}
-		if (lhs.chr == rhs.chr) {
+		if (lhs.chr.getValue() == rhs.chr.getValue()) {
 			return lhs.start < rhs.start && lhs.end < rhs.end && lhs.end >= rhs.start;
 		}
-		return lhs.chr < rhs.chr;
+		return lhs.chr.getValue() < rhs.chr.getValue();
 	}
 
 	bool operator==(const GenomeCoordinates& rhs) const {
-		return ((chr != scaffold && chr == rhs.chr) || (chr == scaffold && chrscaf == rhs.chrscaf)) && start >= rhs.start && end <= rhs.end;
+		return ((!chr.isScaffold() && chr.getValue() == rhs.chr.getValue()) || (chr.isScaffold() && chrscaf == rhs.chrscaf)) && start >= rhs.start && end <= rhs.end;
 	}
 
 	bool operator<(const GenomeCoordinates& rhs) const {
-		if (chr == scaffold && rhs.chr == scaffold && chrscaf == rhs.chrscaf) {
+		if (chr.isScaffold() && rhs.chr.isScaffold() && chrscaf == rhs.chrscaf) {
 			if (start == rhs.start) {
 				return end < rhs.end;
 			}
 			return start < rhs.start;
 		}
-		if (chr == scaffold && rhs.chr == scaffold && chrscaf != rhs.chrscaf) {
+		if (chr.isScaffold() && rhs.chr.isScaffold() && chrscaf != rhs.chrscaf) {
 			return chrscaf < rhs.chrscaf;
 		}
-		if (chr == scaffold && rhs.chr != scaffold) {
+		if (chr.isScaffold() && !rhs.chr.isScaffold()) {
 			return false;
 		}
-		if (chr != scaffold && rhs.chr == scaffold) {
+		if (!chr.isScaffold() && rhs.chr.isScaffold()) {
 			return true;
 		}
-		if (chr == rhs.chr) {
+		if (chr.getValue() == rhs.chr.getValue()) {
 			if (start == rhs.start) {
 				return end < rhs.end;
 			}
 			return start < rhs.start;
 		}
-		return chr < rhs.chr;
+		return chr.getValue() < rhs.chr.getValue();
 	}
 };