Louis BECQUEY / biorseo

Go to a project
Toggle navigation Toggle navigation pinning
Louis BECQUEY
e6293064 1 parent d7ca4934

pattern matching for desc motifs

Inline Side-by-side
Showing 3 changed files with 79 additions and 35 deletions
......@@ -9,7 +9,7 @@ TARGET = biominserter
CC = clang++
# compiling flags here
CFLAGS = -Icppsrc/ -I/usr/local/include -I$(ICONCERT) -I$(ICPLEX) -I$(INUPACK) -I$(IEIGEN) -O3
CFLAGS = -Icppsrc/ -I/usr/local/include -I$(ICONCERT) -I$(ICPLEX) -I$(INUPACK) -I$(IEIGEN) -g
CXXFLAGS = --std=c++17 -Wall -Wpedantic -Wextra -Wno-ignored-attributes -Wno-unused-variable
LINKER = clang++
......@@ -36,6 +36,12 @@ $(OBJECTS): $(OBJDIR)/%.o : $(SRCDIR)/%.cpp $(INCLUDES)
$(CC) -c $(CFLAGS) $(CXXFLAGS) $< -o $@
@echo "\033[00;32mCompiled "$<".\033[00m"
.PHONY: all
all: $(BINDIR)/$(TARGET)
.PHONY: re
re: remove clean all
.PHONY: clean
clean:
$(rm) $(OBJECTS)
......
......@@ -39,19 +39,26 @@ vector<Motif> Motif::build_from_desc(const string& descfile, string rna)
vector<string> bases;
int last;
vector<Motif> results;
char c = 'a';
char* prev = &c;
motif = std::ifstream(descfile);
std::getline(motif, line); // ignore "id: number"
std::getline(motif, line); // Bases: 866_G 867_G 868_G 869_G 870_U 871_A ...
split(bases, line, boost::is_any_of(" ")); // get a vector of 866_G, 867_G, etc...
seq = bases[1].back();
std::getline(motif, line); // ignore "id: number"
std::getline(motif, line); // Bases: 866_G 867_G 868_G 869_G 870_U 871_A ...
boost::split(bases, line, [prev](char c) {
bool res = (*prev == ' ' or *prev == ':');
*prev = c;
return (c == ' ' and res);
}); // get a vector of 866_G, 867_G, etc...
seq = bases[1].substr(bases[1].find('_') + 1, 1);
last = std::stoi(bases[1].substr(0, bases[1].find('_')));
for (vector<string>::iterator b = bases.begin() + 2; b != bases.end(); b++) {
char nt = b->back();
for (vector<string>::iterator b = bases.begin() + 1; b != bases.end() - 1; b++) {
char nt = b->substr(b->find('_') + 1, 1).back();
int pos = std::stoi(b->substr(0, b->find('_')));
if (pos - last > 5) { // finish this component and start a new one
seq += nt;
component_sequences.push_back(seq);
seq = "";
} else if (pos - last == 2) {
......@@ -63,7 +70,7 @@ vector<Motif> Motif::build_from_desc(const string& descfile, string rna)
} else if (pos - last == 5) {
seq += "....";
}
seq += nt;
if (pos - last <= 5) seq += nt;
}
// Now component_sequences is a vector of sequences like {AGCGC, CGU..GUUU}
......@@ -74,6 +81,7 @@ vector<Motif> Motif::build_from_desc(const string& descfile, string rna)
for (vector<Component>& v : vresults) {
results.push_back(Motif(v, descfile.substr(0, descfile.find(".desc"))));
}
std::cout << "\t>returning vector of size " << results.size() << std::endl;
return results;
}
......@@ -115,24 +123,47 @@ vector<vector<Component>> Motif::find_next_ones_in(string rna, vector<string> vc
pair<uint, uint> pos;
vector<vector<Component>> results;
vector<vector<Component>> next_ones;
vector<string> next_seqs(&vc[1], &vc[vc.size() - 1]);
vector<string> next_seqs;
if (vc.size() > 1) {
if (vc.size() > 2)
next_seqs = vector<string>(&vc[1], &vc[vc.size() - 1]);
else
next_seqs = vector<string>(1, vc.back());
std::regex_search(rna, matches, c);
std::regex_search(rna, matches, c);
for (uint i = 0; i < matches.size(); ++i) // Pour chacun des matches
{
pos.first = matches.position(i);
pos.second = matches.length(i) + pos.first - 1;
std::cout << "\t\t>We can insert " << vc[0] << " in [" << pos.first << ',' << pos.second << ']' << std::endl;
std::cout << "\t\t>Now searching in " << rna.substr(pos.second + 1) << std::endl;
next_ones = find_next_ones_in(rna.substr(pos.second + 1), next_seqs);
for (vector<Component> v : next_ones) // Pour chacune des combinaisons suivantes
{
// Combiner le match et la combinaison suivante
vector<Component> r;
r.push_back(Component(pos));
for (Component& c : v) r.push_back(c);
results.push_back(r);
}
}
} else {
std::regex_search(rna, matches, c);
for (uint i = 0; i < matches.size(); ++i) // Pour chacun des matches
{
pos.first = matches.position(i);
pos.second = matches.length(i) + pos.first - 1;
next_ones = find_next_ones_in(rna.substr(pos.second + 1), next_seqs);
for (vector<Component> v : next_ones) // Pour chacune des combinaisons suivantes
for (uint i = 0; i < matches.size(); ++i) // Pour chacun des matches
{
pos.first = matches.position(i);
pos.second = matches.length(i) + pos.first - 1;
std::cout << "\t\t>We can insert " << vc[0] << " in [" << pos.first << ',' << pos.second << ']' << std::endl;
// Combiner le match et la combinaison suivante
vector<Component> r;
r.push_back(Component(pos));
for (Component& c : v) r.push_back(c);
results.push_back(r);
}
}
std::cout << "\t> returning vector of size " << results.size() << std::endl;
return results;
}
......@@ -143,6 +174,8 @@ vector<Motif> load_desc_folder(const string& path, const string& rna, bool verbo
if (!exists(path)) {
std::cerr << "Hmh, i can't find that folder: " << path << std::endl;
return posInsertionSites;
} else {
if (verbose) std::cout << "loading DESC motifs from " << path << "..." << std::endl;
}
for (auto it : recursive_directory_range(path)) {
......@@ -176,16 +209,22 @@ bool is_desc_insertible(const string& descfile, const string& rna, bool verbose)
string seq;
vector<string> bases;
int last;
char c = 'a';
char* prev = &c;
motif = std::ifstream(descfile);
std::getline(motif, line); // ignore "id: number"
std::getline(motif, line); // Bases: 866_G 867_G 868_G 869_G 870_U 871_A ...
split(bases, line, boost::is_any_of(" ")); // get a vector of 866_G, 867_G, etc...
seq = bases[1].back();
std::getline(motif, line); // ignore "id: number"
std::getline(motif, line); // Bases: 866_G 867_G 868_G 869_G 870_U 871_A ...
boost::split(bases, line, [prev](char c) {
bool res = (*prev == ' ' or *prev == ':');
*prev = c;
return (c == ' ' and res);
}); // get a vector of 866_G, 867_G, etc...
seq = "";
last = std::stoi(bases[1].substr(0, bases[1].find('_')));
for (vector<string>::iterator b = bases.begin() + 2; b != bases.end(); b++) {
char nt = b->back();
for (vector<string>::iterator b = (bases.begin() + 1); b != (bases.end() - 1); b++) {
char nt = b->substr(b->find('_') + 1, 1).back();
int pos = std::stoi(b->substr(0, b->find('_')));
if (pos - last > 5) { // finish this component and start a new one
......@@ -206,16 +245,15 @@ bool is_desc_insertible(const string& descfile, const string& rna, bool verbose)
std::regex e(seq);
if (std::regex_search(rna, m, e)) {
if (verbose)
std::cout << "Motif " << descfile.substr(0, descfile.find(".desc")) << " " << seq << " can be inserted." << std::endl;
std::cout << "\t>Motif " << descfile.substr(0, descfile.find(".desc")) << " \t" << seq
<< " can be inserted." << std::endl;
return true;
} else {
if (verbose)
std::cout << "Ignoring motif " << descfile.substr(0, descfile.find(".desc")) << " " << seq << std::endl;
// if (verbose) std::cout << "Ignoring motif " << descfile.substr(0, descfile.find(".desc")) << " \t" << seq << std::endl;
return false;
}
}
bool operator==(const Component& c1, const Component& c2)
{
if (c1.pos.first != c2.pos.first) return false;
......
......@@ -56,14 +56,14 @@ int main(int argc, char* argv[])
if (argc != 6) {
cerr << argc << " arguments specified !" << endl;
cerr << "Please specify the following input files:" << endl;
cerr << "biominserter sequence.fasta insertion.sites.csv prob_threshold verbose obj" << endl;
cerr << "biominserter sequence.fasta motifs_file_or_DESC_folder prob_threshold verbose obj" << endl;
return EXIT_FAILURE;
}
/* VARIABLE DECLARATIONS */
const char* inputName = argv[1];
const char* csvname = argv[2];
const char* motifs_path_name = argv[2];
bool verbose = (atoi(argv[4]) != 0);
string basename = remove_ext(inputName, '.', '/');
float theta_p_threshold = atof(argv[3]);
......@@ -89,13 +89,13 @@ int main(int argc, char* argv[])
if (verbose) cout << "\t>" << inputName << " successfuly loaded (" << myRNA.get_RNA_length() << " nt)" << endl;
// load CSV file
if (access(csvname, F_OK) == -1) {
cerr << csvname << " not found" << endl;
if (access(motifs_path_name, F_OK) == -1) {
cerr << motifs_path_name << " not found" << endl;
return EXIT_FAILURE;
}
posInsertionSites = load_desc_folder(csvname, fa->seq(), verbose);
posInsertionSites = load_desc_folder(motifs_path_name, fa->seq(), verbose);
if (verbose)
cout << "\t>" << csvname << " successfuly loaded (" << posInsertionSites.size() << " insertion sites)" << endl;
cout << "\t>" << motifs_path_name << " successfuly loaded (" << posInsertionSites.size() << " insertion sites)" << endl;
exit(0);
/* FIND PARETO SET */
......