config_enzymes.cpp

Read in the enzymes file.

/*
##############################################################################
# file: config_enzymes.cpp                                                   #
# 'msparser' toolkit example code                                            #
##############################################################################
# COPYRIGHT NOTICE                                                           #
# Copyright 1998-2005 Matrix Science Limited  All Rights Reserved.           #
#                                                                            #
##############################################################################
#    $Source: parser/examples/test_cxx/config_enzymes.cpp $ #
#     $Author: villek@matrixscience.com $                                                      #
#       $Date: 2018-07-30 16:23:53 +0100 $                                         #
#   $Revision: 1b450440f9c97e1e41d0fc6016a27d68951d4532 | MSPARSER_REL_3_0_0-2024-09-24-0-g93ebaeb4f4 $                                                         #
# $NoKeywords::                                                            $ #
##############################################################################
*/
 
#include "msparser.hpp"
 
#include <iostream>
 
 
// All the classes are part of the matrix_science namespace
using namespace matrix_science;
 
int main(int argc, char * argv[])
{
    if ( argc < 2 )
    {
        std::cout << "The location of enzymes file has to be specified as a parameter" << std::endl;
        std::cout << "The location should either be the full path to the enzymes file" << std::endl;
        std::cout << "or a URL to a Mascot server - e.g. http://mascot-server/mascot/cgi" << std::endl;
        return 1;
    }
 
    // A sessionID can optionally be passed as the second parameter
    // This will only be required if the 'file' is a URL
    ms_connection_settings cs;
    if (argc > 2) 
    {
        cs.setSessionID(argv[2]);
    }
    ms_enzymefile file(argv[1], &cs);
 
    if (!file.isValid()) 
    {
        std::cout << "There are errors. Cannot continue. The last error description:" << std::endl;
        std::cout << file.getLastErrorString() << std::endl;
        return 1;
    }
 
    // how many do we have in total?
    int n = file.getNumberOfEnzymes();
    std::cout << "There are " << n << " enzymes definitions available" << std::endl;
 
    // now get them all!
    int i;
    for(i=0; i < n ; i++) 
    {
        const ms_enzyme * enzyme = file.getEnzymeByNumber(i);
        std::cout << enzyme->getTitle() <<  ": ";
        for (int c=0; c < enzyme->getNumberOfCutters(); c++) {
            if (enzyme->getCutterType(c) == matrix_science::ms_enzyme::NTERM_CUTTER) {
                std::cout << "nTerm - ";
            } else {
                std::cout << "cTerm - ";
            }
            std::cout << enzyme->getCleave(c) <<  "!" << enzyme->getRestrict(c);
            std::cout << "; ";
        }
        std::cout << std::endl;
    }
 
    // Now try updating the first one in the list to semi-specific
    ms_enzyme enzyme = *file.getEnzymeByNumber(0);
    enzyme.setSemiSpecific(true);
    file.updateEnzymeByNumber(0, enzyme);
 
    // And delete V8-DE
    if (file.deleteEnzymeByName("V8-DE")) {
        std::cout << "Deleted the enzyme V8-DE " << std::endl;
    }
 
    //Finally, save the file under a new name - but only if not http:
    if (strncmp("http:", argv[1], 5) != 0) {
        std::string filename = argv[1];
        filename +=  ".new";
        file.setFileName(filename.c_str());
        file.save_file();
    }
 
    std::cout << "There are now " << file.getNumberOfEnzymes() << " enzymes definitions available" << std::endl;
    return 0;
}
 
 
 
 
/*
will give the output: 
 
# test.exe ../config/enzymes
 
There are 19 enzymes definitions available
Trypsin: cTerm - KR!P;
Arg-C: cTerm - R!P;
Asp-N: nTerm - BD!;
Asp-N_ambic: nTerm - DE!;
Chymotrypsin: cTerm - FLWY!P;
CNBr: cTerm - M!;
CNBr+Trypsin: cTerm - M!; cTerm - KR!P;
Formic_acid: cTerm - D!;
Lys-C: cTerm - K!P;
Lys-C/P: cTerm - K!;
PepsinA: cTerm - FL!;
Tryp-CNBr: cTerm - KMR!P;
TrypChymo: cTerm - FKLRWY!P;
Trypsin/P: cTerm - KR!;
V8-DE: cTerm - BDEZ!P;
V8-E: cTerm - EZ!P;
semiTrypsin: cTerm - KR!P;
LysC+AspN: nTerm - BD!; cTerm - K!P;
None: cTerm - KR!P;
There are now 18 enzymes definitions available
 
*/