/// \brief Load a tree and a collection of alignments based on command line parameters.
///
/// \param args The command line parameters.
/// \param alignments The alignments.
/// \param T The leaf-labelled tree.
/// \param internal_sequences Should each resulting alignment have sequences for internal nodes on the tree?
///
void load_As_and_T(const variables_map& args,vector<alignment>& alignments,RootedSequenceTree& T,const vector<bool>& internal_sequences)
{
alignments = load_As(args);
T = load_T(args);
link(alignments,T,internal_sequences);
for(int i=0;i<alignments.size();i++)
{
//---------------- Randomize alignment? -----------------//
if (args.count("randomize-alignment"))
alignments[i] = randomize(alignments[i],T.n_leaves());
//------------------ Analyze 'internal'------------------//
if ((args.count("internal") and args["internal"].as<string>() == "+")
or args.count("randomize-alignment"))
for(int column=0;column< alignments[i].length();column++) {
for(int j=T.n_leaves();j<alignments[i].n_sequences();j++)
alignments[i](column,j) = alphabet::not_gap;
}
//---- Check that internal sequence satisfy constraints ----//
check_alignment(alignments[i],T,internal_sequences[i]);
}
}
/// \brief Load a tree and an alignment based on command line parameters.
///
/// \param args The command line parameters.
/// \param alignments The alignments.
/// \param T The leaf-labelled tree.
/// \param internal_sequences Should each resulting alignment have sequences for internal nodes on the tree?
///
void load_A_and_T(const variables_map& args,alignment& A,RootedSequenceTree& T,bool internal_sequences)
{
A = load_A(args,internal_sequences);
T = load_T(args);
//------------- Link Alignment and Tree -----------------//
link(A,T,internal_sequences);
//---------------- Randomize alignment? -----------------//
if (args.count("randomize-alignment"))
A = randomize(A,T.n_leaves());
else if (args.count("unalign-all"))
A = unalign_all(A,T.n_leaves());
//------------------ Analyze 'internal'------------------//
if ((args.count("internal") and args["internal"].as<string>() == "+")
or args.count("randomize-alignment"))
for(int column=0;column< A.length();column++) {
for(int i=T.n_leaves();i<A.n_sequences();i++)
A.set_value(column,i, alphabet::not_gap );
}
//---- Check that internal sequence satisfy constraints ----//
check_alignment(A,T,internal_sequences);
}
/// Load a tree from command line args "--tree filename"
RootedSequenceTree load_T(const variables_map& args) {
if (not args.count("tree"))
throw myexception()<<"Tree file not specified! (--tree <filename>)";
RootedSequenceTree RT;
RT.read(args["tree"].as<string>());
return RT;
}
RootedSequenceTree::RootedSequenceTree(const RootedSequenceTree& T1, const RootedSequenceTree& T2)
:RootedTree(T1,T2)
{
// We will create new names which will be the same as
// T1.order + T2.order
for(int i=0;i<T1.get_sequences().size();i++)
sequences.push_back(T1.seq(i));
for(int i=0;i<T2.get_sequences().size();i++)
sequences.push_back(T2.seq(i));
}
/// Construct a multifurcating tree representing topology constraints from file \a filename.
///
/// \param filename The name of the file to load the tree from.
/// \param names The order of the leaf labels.
/// \return a multifurcating tree.
///
SequenceTree load_constraint_tree(const string& filename,const vector<string>& names)
{
RootedSequenceTree RT;
RT.read(filename);
SequenceTree constraint = RT;
remove_sub_branches(constraint);
try{
remap_T_indices(constraint,names);
}
catch(const bad_mapping<string>& b) {
bad_mapping<string> b2(b.missing,b.from);
if (b.from == 0)
b2<<"Constraint tree leaf sequence '"<<b2.missing<<"' not found in the alignment.";
else
b2<<"Alignment sequence '"<<b2.missing<<"' not found in the constraint tree.";
throw b2;
}
return constraint;
}
vector<SequenceTree> load_trees(const vector<string>& lines)
{
if (lines.size() == 0)
throw myexception()<<"No trees were read in!";
vector<SequenceTree> trees;
for(int i=0;i<lines.size();i++)
{
RootedSequenceTree T;
try {
T.parse(lines[i]);
}
catch (std::exception& e) {
cerr<<"Exception: "<<e.what()<<endl;
cerr<<" Quitting read of tree file"<<endl;
break;
}
trees.push_back(T);
}
return trees;
}
/// \brief Remap the leaf indices of tree \a T to match the alignment \a A: check the result
///
/// \param A The alignment.
/// \param T The tree.
/// \param internal_sequences Should the resulting alignment have sequences for internal nodes on the tree?
///
void link(alignment& A,RootedSequenceTree& T,bool internal_sequences)
{
check_names_unique(A);
// Later, might we WANT sub-branches???
if (has_sub_branches(T))
remove_sub_branches(T);
if (internal_sequences and not is_Cayley(T)) {
assert(has_polytomy(T));
throw myexception()<<"Cannot link a multifurcating tree to an alignment with internal sequences.";
}
//------ IF sequences < leaf nodes THEN complain ---------//
if (A.n_sequences() < T.n_leaves())
throw myexception()<<"Tree has "<<T.n_leaves()<<" leaves but Alignment only has "
<<A.n_sequences()<<" sequences.";
//----- IF sequences = leaf nodes THEN maybe add internal sequences.
else if (A.n_sequences() == T.n_leaves()) {
if (internal_sequences)
A = add_internal(A,T);
}
//----- IF sequences > leaf nodes THEN maybe complain -------//
else {
if (not internal_sequences)
throw myexception()<<"More alignment sequences than leaf nodes!";
if (A.n_sequences() > T.n_nodes())
throw myexception()<<"More alignment sequences than tree nodes!";
else if (A.n_sequences() < T.n_nodes())
throw myexception()<<"Fewer alignment sequences than tree nodes!";
}
//---------- double-check that we have the right number of sequences ---------//
if (internal_sequences)
assert(A.n_sequences() == T.n_nodes());
else
assert(A.n_sequences() == T.n_leaves());
//----- Remap leaf indices for T onto A's leaf sequence indices -----//
remap_T_indices(T,A);
if (internal_sequences)
connect_leaf_characters(A,T);
//---- Check to see that internal nodes satisfy constraints ----//
check_alignment(A,T,internal_sequences);
}
int main(int argc,char* argv[])
{
try {
//---------- Parse command line -------//
variables_map args = parse_cmd_line(argc,argv);
RootedSequenceTree T = load_T(args);
int root=-1;
if (args.count("outgroup"))
{
string outgroup = args["outgroup"].as<string>();
int leaf = find_leaf(T,outgroup);
root = split_branch(T,leaf);
}
else if (args.count("taxa"))
{
string taxa = args["taxa"].as<string>();
vector<string> taxon = split(taxa,',');
if (taxon.size() != 3)
throw myexception()<<"You must supply exactly 3 taxa, but you supplied "<<taxon.size();
int n1 = find_leaf(T,taxon[0]);
int n2 = find_leaf(T,taxon[1]);
int n3 = find_leaf(T,taxon[2]);
T.reroot(n1);
root = T.common_ancestor(n2,n3);
}
else if (args.count("leaf"))
{
string leaf_name = args["leaf"].as<string>();
root = find_leaf(T,leaf_name);
}
else if (args.count("parent-of"))
{
string leaf_name = args["parent-of"].as<string>();
int leaf = find_leaf(T,leaf_name);
root = T.branch(leaf).target();
}
else if (args.count("branch"))
{
string p = args["branch"].as<string>();
vector<string> taxa = split(p,' ');
dynamic_bitset<> mask(T.n_leaves());
dynamic_bitset<> group1(T.n_leaves());
int separator = find_index(taxa,string("|"));
if (separator == -1)
throw myexception()<<"Partition is missing a separator";
for(int i=0;i<separator;i++)
{
int ii = find_leaf(T,taxa[i]);
mask[ii] = true;
group1[ii] = true;
}
for(int i=separator+1;i<taxa.size();i++)
{
int ii = find_leaf(T,taxa[i]);
mask[ii] = true;
}
Partition P(T.get_sequences(),group1,mask);
cerr<<P<<endl;
int b = which_partition(T,P);
cerr<<partition_from_branch(T,b)<<endl;
root = split_branch(T,b);
}
else
throw myexception("neither --outgroup nor --taxa nor --leaf specified!");
T.reroot(root);
std::cout<<T<<endl;
}
catch (std::exception& e) {
std::cerr<<"tree-reroot: Error! "<<e.what()<<endl;
exit(1);
}
return 0;
}
bool reader_t::next_tree(RootedSequenceTree& T)
{
T.get_sequences() = leaf_names;
return next_tree(static_cast<RootedTree&>(T));
}