From 15aeecde4dc2276f523fa4e126f144b822dbf747 Mon Sep 17 00:00:00 2001
From: Dariusz Murakowski <murakdar@mit.edu>
Date: Thu, 23 Apr 2015 23:10:32 -0400
Subject: [PATCH] Start of Potts / amino acid / nucleotide-based virus species.

---
 pop_ss.cpp | 33 +++++++++++++++++++++++++++++++++
 pop_ss.h   | 32 ++++++++++++++++++++++++++++++++
 ss.cpp     |  7 ++++++-
 3 files changed, 71 insertions(+), 1 deletion(-)

diff --git a/pop_ss.cpp b/pop_ss.cpp
index 05ff5e7..fb4851b 100644
--- a/pop_ss.cpp
+++ b/pop_ss.cpp
@@ -36,6 +36,39 @@ bool operator<(const Virus& lhs, const Virus& rhs) {
     return lhs.mutated_sites < rhs.mutated_sites;
 }
 
+std::vector<aa> translate_nt_seq(const std::string &nt_seq)
+{
+    std::vector<aa> aa_seq;  aa_seq.reserve(nt_seq.size() / 3);
+    for (unsigned i=0; i<nt_seq.length(); i+=3) {
+        aa_seq.push_back(codon2aa.at(str2codon.at(nt_seq.substr(i,3))));
+    }
+    return aa_seq;
+}
+
+std::string aaseq2str(const std::vector<aa> &aa_seq)
+{
+    std::string out;  out.reserve(aa_seq.size());
+    for (std::vector<aa>::const_iterator it  = aa_seq.begin(),
+                                         end = aa_seq.end();
+            it != end; ++it) {
+        out.push_back(aa2str.at(*it));
+    }
+    return out;
+}
+
+NTVirus::NTVirus(const std::string &NT)
+: energy(0.0)
+, nt_seq(NT)
+, aa_seq(translate_nt_seq(NT))
+, L_nt(NT.length())
+, L_aa(this->L_nt/3)
+{
+}
+
+bool operator<(const NTVirus& lhs, const NTVirus& rhs) {
+    return lhs.nt_seq < rhs.nt_seq;
+}
+
 
 // Did virus escape from immune pressure?
 bool EpitopeRecognizer::escaped(const Virus &v) const {
diff --git a/pop_ss.h b/pop_ss.h
index 26cf280..5b02bd9 100644
--- a/pop_ss.h
+++ b/pop_ss.h
@@ -8,6 +8,7 @@
 #include <gsl/gsl_rng.h>
 
 #include "ham_ss.h"
+#include "seqTools.h"
 
 
 class Hamiltonian;
@@ -40,6 +41,28 @@ typedef std::map<Virus, unsigned int> virus_map;
 typedef std::set<unsigned int> MutatedSiteSequence;
 
 
+// helper functions
+std::vector<aa> translate_nt_seq(const std::string &nt_seq);
+std::string aaseq2str(const std::vector<aa> &aa_seq);
+
+class NTVirus {
+public:
+    double energy;
+    std::string nt_seq;
+    std::vector<aa> aa_seq;
+    unsigned int L_nt;
+    unsigned int L_aa;
+
+    NTVirus(const std::string &);
+};
+
+// needed for STL map
+bool operator<(const NTVirus& lhs, const NTVirus& rhs);
+
+typedef std::map<NTVirus, unsigned int> NTVirus_map;
+
+
+
 class EpitopeRecognizer {
 public:
     std::vector<unsigned int> epitopeWT;  // sites with state=0 are recognized
@@ -80,6 +103,15 @@ public:
     virus_map pop;
 };
 
+class NTVirusSpecies : public Species {
+public:
+    NTVirusSpecies() : Species() {}
+    NTVirusSpecies(long c) : Species(c) {}
+    NTVirusSpecies(std::string x) : Species(x) {}
+    NTVirusSpecies(std::string x, long c) : Species(x,c) {}
+    NTVirus_map pop;
+};
+
 class CTLSpecies : public Species {
 public:
     CTLSpecies() : Species(), num_ep(0) {}
diff --git a/ss.cpp b/ss.cpp
index 94dc750..115ad06 100644
--- a/ss.cpp
+++ b/ss.cpp
@@ -497,7 +497,12 @@ int main(int argc, char *argv[]) {
         }
         std::cout << '\n';
     }
-    std::cout << int2aa.at(codon2aa.at(str2codon.at("TTT"))) << '\n';
+    //NTVirus v("TTT");
+    NTVirus v(argv[2]);
+    //std::cout << int2aa.at(codon2aa.at(str2codon.at(v.nt_seq))) << '\n';
+    std::cout << aa2str.at(v.aa_seq[0]) << '\n';
+    std::cout << aaseq2str(v.aa_seq) << '\n';
+    printf("%d %d\n",v.L_nt,v.L_aa);
     return 0;
 
     RunParameters_SS r;
-- 
2.7.4