void MotifSearch::modify_params(int argc, char *argv[]){
GetArg2(argc, argv, "-expect", params.expect);
GetArg2(argc, argv, "-minpass", params.minpass);
GetArg2(argc, argv, "-seed", params.seed);
GetArg2(argc, argv, "-undersample", params.undersample);
GetArg2(argc, argv, "-oversample", params.oversample);
}
JBoolean
JBooleanXOR::Evaluate()
const
{
const JBoolean arg1 = (GetArg1())->Evaluate();
const JBoolean arg2 = (GetArg2())->Evaluate();
return JConvertToBoolean((arg1 && !arg2) || (!arg1 && arg2));
}
void dCILInstrConditional::SetTargets (dCILInstrLabel* const target0, dCILInstrLabel* const target1)
{
dAssert(target0);
dAssert (target0->GetArg0().m_label == GetArg1().m_label);
m_tagetNode0 = target0->GetNode();
if (target1) {
dAssert(target1->GetArg0().m_label == GetArg2().m_label);
m_tagetNode1 = target1->GetNode();
} else {
m_tagetNode1 = NULL;
}
}
JBoolean
JLogB::Evaluate
(
JFloat* result
)
const
{
JFloat b,x;
if (!(GetArg1())->Evaluate(&b) || b <= 1.0)
{
return kJFalse;
}
if (!(GetArg2())->Evaluate(&x))
{
return kJFalse;
}
jclear_errno();
*result = log(x)/log(b);
return jerrno_is_clear();
}
void
JLogB::Render
(
const JExprRenderer& renderer,
const JExprRectList& rectList
)
const
{
// find ourselves in the list
JIndex ourIndex;
const JBoolean found = rectList.FindFunction(this, &ourIndex);
assert( found );
const JRect ourRect = rectList.GetRect(ourIndex);
const JCoordinate ourMidline = rectList.GetMidline(ourIndex);
const JSize fontSize = rectList.GetFontSize(ourIndex);
// draw ourselves
const JCharacter* name = GetName();
const JSize nameLength = strlen(name);
assert( nameLength > 1 );
assert( name[ nameLength-1 ] == '(' );
const JString fnName = JString(name, nameLength-1);
renderer.DrawString(ourRect.left, ourMidline, fontSize, fnName);
// draw the base
(GetArg1())->Render(renderer, rectList);
// draw the argument
const JFunction* arg = GetArg2();
arg->Render(renderer, rectList);
JIndex argIndex;
const JBoolean foundArg = rectList.FindFunction(arg, &argIndex);
assert( foundArg );
renderer.DrawParentheses(rectList.GetRect(argIndex));
}
JBoolean
JBooleanAND::Evaluate()
const
{
return JConvertToBoolean( (GetArg1())->Evaluate() && (GetArg2())->Evaluate() );
}
JIndex
JLogB::PrepareToRender
(
const JExprRenderer& renderer,
const JPoint& upperLeft,
const JSize fontSize,
JExprRectList* rectList
)
{
// intialize our rectangle
const JCharacter* name = GetName();
const JSize nameLength = strlen(name);
assert( nameLength > 1 );
assert( name[ nameLength-1 ] == '(' );
const JString fnName = JString(name, nameLength-1);
JRect ourRect;
ourRect.top = upperLeft.y;
ourRect.left = upperLeft.x;
ourRect.bottom = upperLeft.y + renderer.GetLineHeight(fontSize);
ourRect.right = upperLeft.x + renderer.GetStringWidth(fontSize, fnName);
// get rectangle for base
JPoint argUpperLeft(ourRect.right, ourRect.top);
const JSize baseFontSize = renderer.GetSuperSubFontSize(fontSize);
JFunction* base = GetArg1();
const JIndex baseIndex =
base->PrepareToRender(renderer, argUpperLeft, baseFontSize, rectList);
const JRect baseRect = rectList->GetRect(baseIndex);
argUpperLeft.x = baseRect.right;
// get rectangle for argument -- gives our midline
JFunction* arg = GetArg2();
const JIndex argIndex =
arg->PrepareToRender(renderer, argUpperLeft, fontSize, rectList);
const JRect argRect = rectList->GetRect(argIndex);
ourRect = JCovering(ourRect, argRect);
const JCoordinate ourMidline = rectList->GetMidline(argIndex);
// shift argument to make space for left parenthesis
const JSize parenWidth = renderer.GetParenthesisWidth(argRect.height());
rectList->ShiftRect(argIndex, parenWidth, 0);
// we need space for two parentheses
ourRect.right += 2*parenWidth;
// shift the base down
rectList->ShiftRect(baseIndex, 0, ourMidline - ourRect.top);
ourRect = JCovering(ourRect, rectList->GetRect(baseIndex));
// save our rectangle
return rectList->AddRect(ourRect, ourMidline, fontSize, this);
}
int main(int argc, char *argv[]) {
set_new_handler(alloc_error);
if(argc < 6) {
print_usage(cout);
exit(0);
}
string seqfile; // file with sequences
string exprfile; // file with expression data
string subsetfile; // file with subset of sequence names to search
string scorefile; // file with scores
if(! GetArg2(argc, argv, "-s", seqfile)) {
cerr << "Please specify sequence file\n\n";
print_usage(cout);
exit(0);
}
if(! GetArg2(argc, argv, "-o", outfile)) {
cerr << "Please specify output file\n\n";
print_usage(cout);
exit(0);
}
int search_type = UNDEFINED;
if(GetArg2(argc, argv, "-ex", exprfile)) {
search_type = EXPRESSION;
}
if(GetArg2(argc, argv, "-su", subsetfile)) {
search_type = SUBSET;
}
if(GetArg2(argc, argv, "-sc", scorefile)) {
search_type = SCORE;
}
if(search_type == UNDEFINED) {
cerr << "Please specify either an expression data file, a file with a subset of sequence names, or a file with sequence scores.\n\n";
print_usage(cout);
exit(0);
}
// Decide mode of running
archive = false;
if(! GetArg2(argc, argv, "-worker", worker)) {
worker = -1;
archive = true;
}
// Read parameters
vector<string> seqs;
cerr << "Reading sequence data from '" << seqfile << "'... ";
get_fasta_fast(seqfile.c_str(), seqs, seq_nameset);
cerr << "done.\n";
ngenes = seq_nameset.size();
npoints = 0;
if(search_type == EXPRESSION) {
cerr << "Reading expression data from '" << exprfile << "'... ";
get_expr(exprfile.c_str(), expr, data_nameset);
cerr << "done.\n";
npoints = expr[0].size();
nsubset = 0;
} else if(search_type == SUBSET) {
cerr << "Reading subset of sequences to search from '" << subsetfile << "'... ";
get_list(subsetfile.c_str(), subset);
cerr << "done.\n";
npoints = 0;
for(int i = 0; i < ngenes; i++) {
vector<float> row(0);
expr.push_back(row);
}
nsubset = subset.size();
sort(subset.begin(), subset.end());
} else if(search_type == SCORE) {
cerr << "Reading sequence scores from '" << scorefile << "'... ";
get_scores(scorefile.c_str(), scores, data_nameset);
cerr << "done.\n";
npoints = 1;
nsubset = 0;
}
vector<vector <float> > newexpr;
vector<float> newscores;
if(search_type == EXPRESSION) {
order_data_expr(newexpr);
}
if(search_type == SCORE) {
order_data_scores(newscores);
}
if(search_type == EXPRESSION) {
cerr << "Successfully read input files -- dataset size is " << ngenes << " sequences X " << npoints << " timepoints\n";
} else if(search_type == SUBSET) {
cerr << "Successfully read input files -- dataset size is " << ngenes << " sequences, with " << nsubset << " to be searched\n";
} else if(search_type == SCORE) {
cerr << "Successfully read input files -- dataset size is " << ngenes << " scored sequences\n";
}
cerr << "Setting up MotifSearch... ";
if(! GetArg2(argc, argv, "-numcols", ncol)) ncol = 10;
if(! GetArg2(argc, argv, "-order", order)) order = 0;
if(! GetArg2(argc, argv, "-simcut", simcut)) simcut = 0.8;
if(! GetArg2(argc, argv, "-maxm", maxm)) maxm = 20;
MotifSearch* ms;
if(search_type == EXPRESSION) {
ms = new MotifSearchExpr(seq_nameset, seqs, ncol, order, simcut, maxm, newexpr, npoints);
} else if(search_type == SCORE) {
ms = new MotifSearchScore(seq_nameset, seqs, ncol, order, simcut, maxm, newscores);
} else {
ms = new MotifSearchSubset(seq_nameset, seqs, ncol, order, simcut, maxm, subset);
}
ms->modify_params(argc, argv);
ms->set_final_params();
ms->ace_initialize();
cerr << "done.\n";
cerr << "Random seed: " << ms->get_params().seed << '\n';
if(archive) {
cerr << "Running in archive mode...\n";
string archinstr(outfile);
archinstr.append(".ms");
ifstream archin(archinstr.c_str());
if(archin) {
cerr << "Refreshing from existing archive file " << archinstr << "... ";
ms->get_archive().read(archin);
cerr << "done.\n";
}
while(true) {
int found = read_motifs(ms);
if(found > 0) output(ms);
if(found < 50) sleep(10);
}
} else {
cerr << "Running as worker " << worker << "...\n";
int nruns = ms->positions_in_search_space()/(ms->get_params().expect * ncol);
nruns *= ms->get_params().oversample;
nruns /= ms->get_params().undersample;
cerr << "Restarts planned: " << nruns << '\n';
string archinstr(outfile);
archinstr.append(".ms");
string lockstr(outfile);
lockstr.append(".lock");
for(int j = 1; j <= nruns; j++) {
if(j == 1 || j % 50 == 0 || search_type == SUBSET) {
struct flock fl;
int fd;
fl.l_type = F_RDLCK;
fl.l_whence = SEEK_SET;
fl.l_start = 0;
fl.l_len = 0;
fl.l_pid = getpid();
fd = open(lockstr.c_str(), O_RDONLY);
if(fd == -1) {
if(errno != ENOENT)
cerr << "\t\tUnable to read lock file, error was " << strerror(errno) << "\n";
} else {
while(fcntl(fd, F_SETLK, &fl) == -1) {
cerr << "\t\tWaiting for lock release on archive file... \n";
sleep(10);
}
ifstream archin(archinstr.c_str());
if(archin) {
cerr << "\t\tRefreshing archive from " << archinstr << "...";
ms->get_archive().clear();
ms->get_archive().read(archin);
archin.close();
cerr << "done.\n";
}
fl.l_type = F_UNLCK;
fcntl(fd, F_SETLK, &fl);
close(fd);
cerr << "\t\tArchive now has " << ms->get_archive().nmots() << " motifs\n";
}
}
cerr << "\t\tSearch restart #" << j << "/" << nruns << "\n";
ms->search_for_motif(worker, j, outfile);
}
}
delete ms;
return 0;
}
