restriction_sites.cpp

来自「ncbi源码」· C++ 代码 · 共 651 行 · 第 1/2 页

            // do the big search
            CFindRSites::Find(vec, enzymes, results);

            // add description to annot
            annot->SetName("Restriction sites");

            // deal with stored enzyme results

            sort(results.begin(), results.end(), SEnzymeNameCompare());

            if (args["sort_by_number_of_sites"].AsBoolean()) {
                // do a stablesort so alphabetical order preserved
                // within sets with the same number of sites
                stable_sort(results.begin(), results.end(), SLessDefSites());
            }

            // count the number of definite and possible sites
            int total_definite_sites = 0, total_possible_sites = 0;
            int total_non_cutters = 0;
            ITERATE (TResults, result, results) {
                total_definite_sites += (*result)->GetDefiniteSites().size();
                total_possible_sites += (*result)->GetPossibleSites().size();
                if ((*result)->GetDefiniteSites().size()
                    + (*result)->GetPossibleSites().size() == 0) {
                    total_non_cutters++;
                }
            }
            _TRACE("Found " << total_definite_sites << " definite and "
                   << total_possible_sites << " possible sites");

            // in order to build dialog efficiently,
            // pre-allocate rows
            int total_rows = total_definite_sites + total_possible_sites
                + total_non_cutters;
            m_Dialog->SetRows(row + total_rows);

            ITERATE (TResults, result, results) {
                
                //
                // add sites to dialog
                //

                int starting_row = row;
                m_Dialog->SetCell(row, 2) = (*result)->GetEnzymeName();
                string num_sites =
                    NStr::IntToString((*result)->GetDefiniteSites().size());
                if (!(*result)->GetPossibleSites().empty()) {
                    num_sites += "+";
                    num_sites +=
                        NStr::IntToString((*result)
                                          ->GetPossibleSites().size());
                    num_sites += "?";
                }
                m_Dialog->SetCell(row, 3) = num_sites;

                const vector<CRSite>& definite_sites
                    = (*result)->GetDefiniteSites();
                ITERATE (vector<CRSite>, site, definite_sites) {
                    string& pos_str = m_Dialog->SetCell(row, 4);
                    pos_str = NStr::IntToString(site->GetStart() + 1) + " - "
                        + NStr::IntToString(site->GetEnd() + 1);

                    string plus_cuts;
                    const vector<int>& pcuts = site->GetPlusCuts();
                    ITERATE (vector<int>, cut, pcuts) {
                        if (!plus_cuts.empty()) {
                            plus_cuts += ", ";
                        }
                        plus_cuts += NStr::IntToString(*cut);
                    }
                    m_Dialog->SetCell(row, 5) = plus_cuts;

                    string minus_cuts;
                    const vector<int>& mcuts = site->GetMinusCuts();
                    ITERATE (vector<int>, cut, mcuts) {
                        if (!minus_cuts.empty()) {
                            minus_cuts += ", ";
                        }
                        minus_cuts += NStr::IntToString(*cut);
                    }
                    m_Dialog->SetCell(row, 6) = minus_cuts;

                    ++row;
                }

                const vector<CRSite>& possible_sites
                    = (*result)->GetPossibleSites();
                ITERATE (vector<CRSite>, site, possible_sites) {
                    string& pos_str = m_Dialog->SetCell(row, 4);
                    pos_str = string("???")
                        + NStr::IntToString(site->GetStart() + 1) + " - "
                        + NStr::IntToString(site->GetEnd() + 1);

                    string plus_cuts;
                    const vector<int>& pcuts = site->GetPlusCuts();
                    ITERATE (vector<int>, cut, pcuts) {
                        if (!plus_cuts.empty()) {
                            plus_cuts += ", ";
                        }
                        plus_cuts += NStr::IntToString(*cut);
                    }
                    m_Dialog->SetCell(row, 5) = plus_cuts;

                    string minus_cuts;
                    const vector<int>& mcuts = site->GetMinusCuts();
                    ITERATE (vector<int>, cut, mcuts) {
                        if (!minus_cuts.empty()) {
                            minus_cuts += ", ";
                        }
                        minus_cuts += NStr::IntToString(*cut);
                    }
                    m_Dialog->SetCell(row, 6) = minus_cuts;

                    ++row;
                }

                if (row == starting_row) {
                    row++;
                }

                //
                // add features to annot
                //
                s_AddSitesToAnnot(definite_sites, **result, *annot, loc);
                s_AddSitesToAnnot(possible_sites, **result, *annot, loc, false);
            }

            // attach annot to doc
            //const_cast<IDocument&>(doc).AttachAnnot(*annot);
            //const_cast<IDocument&>(doc).UpdateAllViews(fDocumentChanged);
            reply.AddObject(doc, *annot);
        }
        catch (exception& e) {
            LOG_POST(Error << e.what());
            string str = CPluginUtils::GetLabel(loc, &doc.GetScope());
            LOG_POST(Error << "Error processing location " << str);
        }
#ifndef _DEBUG
        catch (...) {
            string str = CPluginUtils::GetLabel(loc, &doc.GetScope());
            LOG_POST(Error << "Error processing location " << str);
        }
#endif
    }

    // update all views
    //CDocManager::UpdateAllViews();

    //
    // prepare our dialog box
    //
    m_Dialog->SetLabel(string("A search for restriction sites for ")
                       + NStr::IntToString(enzymes.size())
                       + " enzymes produced:");
    m_Dialog->Show();
    reply.SetStatus(eMessageStatus_success);
    reply.AddAction(CPluginReplyAction::e_Add_to_document);
}


void 
CAlgoPlugin_RestrictionSites::x_LoadREnzymeData(vector<CREnzyme>& enzymes,
                                     CRebase::EEnzymesToLoad which_enzymes)
{
    CNcbiApplication* app = CNcbiApplication::Instance();
    _ASSERT(app);
    CNcbiRegistry& registry = app->GetConfig();
    string fname;

    // By default the rebase "NAR format" file 
    // is assumed to be <std>/etc/rebase.nar.
    // This can be overridden in gbench.ini, via the application registry
    // variable [RESTRICTION_SITES] RebaseData.
    fname = registry.GetString("RESTRICTION_SITES", "RebaseData", "");
    if ( !fname.empty() ) {
        fname += ", ";
    }
    fname += "<home>/etc/rebase.nar, <std>/etc/rebase.nar";
    fname = CSystemPath::ResolvePathExisting(fname);
    if ( fname.empty() ) {
        throw runtime_error("Couldn't open REBASE file ");
    }

    ifstream refile(fname.c_str());
    CRebase::ReadNARFormat(refile, enzymes, which_enzymes);
}


// helper function to decode a parameter
CRebase::EEnzymesToLoad
CAlgoPlugin_RestrictionSites::x_DecodeWhichEnzymes(const string& s)
{
    if (s == "commercially available") {
        return CRebase::eCommercial;
    } else if (s == "prototypes") {
        return CRebase::ePrototype;
    } else if (s == "all") {
        return CRebase::eAll;
    } else {
        throw runtime_error(string("Invalid \"which_enzymes\" string: ")
                            + s);
    }
}


END_NCBI_SCOPE


/*
 * ===========================================================================
 * $Log: restriction_sites.cpp,v $
 * Revision 1000.5  2004/06/01 20:55:43  gouriano
 * PRODUCTION: UPGRADED [GCC34_MSVC7] Dev-tree R1.32
 *
 * Revision 1.32  2004/05/21 22:27:47  gorelenk
 * Added PCH ncbi_pch.hpp
 *
 * Revision 1.31  2004/05/03 13:05:42  dicuccio
 * gui/utils --> gui/objutils where needed
 *
 * Revision 1.30  2004/03/05 17:35:37  dicuccio
 * Use sequence::GetId() instead of CSeq_id::GetStringDescr()
 *
 * Revision 1.29  2004/02/17 20:35:25  rsmith
 * moved core/settings.[ch]pp and core/system_path.[ch]pp to config and utils, respectively.
 *
 * Revision 1.28  2004/01/27 18:38:09  dicuccio
 * Code clean-up.  Use standard names for plugins.  Removed unnecessary #includes
 *
 * Revision 1.27  2004/01/07 15:50:39  dicuccio
 * Adjusted for API change in CPluginUtils::GetLabel().  Standardized exception
 * reporting in algorithms.
 *
 * Revision 1.26  2003/12/31 15:36:08  grichenk
 * Moved CompareLocations() from CSeq_feat to CSeq_loc,
 * renamed it to Compare().
 *
 * Revision 1.25  2003/11/24 15:45:28  dicuccio
 * Renamed CVersion to CPluginVersion
 *
 * Revision 1.24  2003/11/18 17:48:38  dicuccio
 * Added standard processing of return values
 *
 * Revision 1.23  2003/11/04 17:49:23  dicuccio
 * Changed calling parameters for plugins - pass CPluginMessage instead of paired
 * CPluginCommand/CPluginReply
 *
 * Revision 1.22  2003/10/27 17:46:49  dicuccio
 * Removed dead #includes
 *
 * Revision 1.21  2003/10/17 19:10:42  dicuccio
 * Code clean-up.  Refactored feature building into a more modular function.
 * Fixed list sorting - replaced with sorting of a vector to work around lack of
 * support for list::sort(functor) in MSVC
 *
 * Revision 1.20  2003/10/17 18:07:47  jcherry
 * Sort Seq-locs in mix by location.  Remap features for possible sites too.
 *
 * Revision 1.19  2003/10/15 13:40:26  dicuccio
 * Mkae sure to set the 'id' for the seq-locs before calling RemapChildToParent()
 *
 * Revision 1.18  2003/10/14 16:24:02  dicuccio
 * Added correct remapping of scanned locations to the parent location.  Cleaned
 * up code to look for data file - added hierarchichal search through path in INI,
 * user's home directory, and finally system installed path.
 *
 * Revision 1.17  2003/10/07 13:47:00  dicuccio
 * Renamed CPluginURL* to CPluginValue*
 *
 * Revision 1.16  2003/09/25 17:21:35  jcherry
 * Added name to annot
 *
 * Revision 1.15  2003/09/04 14:05:24  dicuccio
 * Use IDocument instead of CDocument
 *
 * Revision 1.14  2003/09/03 14:46:53  rsmith
 * change namespace name from args to plugin_args to avoid clashes with variable names.
 *
 * Revision 1.13  2003/08/22 15:47:45  dicuccio
 * Added 'objects::' where necessary
 *
 * Revision 1.12  2003/08/21 19:24:16  jcherry
 * Moved restriction site finding to algo/sequence
 *
 * Revision 1.11  2003/08/21 18:38:32  jcherry
 * Overloaded CFindRSites::Find to take several sequence containers.
 * Added option to lump together enzymes with identical specificities.
 *
 * Revision 1.10  2003/08/21 12:03:07  dicuccio
 * Make use of new typedef in plugin_utils.hpp for argument values.
 *
 * Revision 1.9  2003/08/20 22:57:44  jcherry
 * Reimplemented restriction site finding using finite state machine
 *
 * Revision 1.8  2003/08/18 19:24:15  jcherry
 * Moved orf and seq_match to algo/sequence
 *
 * Revision 1.7  2003/08/15 16:57:17  jcherry
 * For consecutive enzymes with identical specificities, reuse
 * search results.  This saves a bunch of time.
 *
 * Revision 1.6  2003/08/15 16:33:40  jcherry
 * Fixed typo
 *
 * Revision 1.5  2003/08/15 15:26:12  jcherry
 * Changed so that restriction site searching (CFindRSites::Find) returns
 * a vector of CRefs rather than a vector of objects.  This speeds sorting.
 *
 * Revision 1.4  2003/08/14 18:33:32  jcherry
 * Use an rsite feature type (rather than a region).
 * Also include cleavage sites in feature location.
 *
 * Revision 1.3  2003/08/13 17:40:26  dicuccio
 * Formatting fixes.  Changes some pass-by-val to pass-by-reference.  Fixed
 * complement table
 *
 * Revision 1.2  2003/08/13 12:37:58  dicuccio
 * Partial compilation fixes for Windows
 *
 * Revision 1.1  2003/08/12 18:52:58  jcherry
 * Initial version
 *
 * ===========================================================================
 */