proteome/proteome.h

#ifndef _PROTEOME_H_
#define _PROTEOME_H_

extern           char* PROTEIN_DATABASE_PATH            ;
extern const char* PROTEIN_DATABASE_FOLDER              ;
extern const char* PROTEIN_DATABASE_INDEX_FILE  ;

extern           char* PROTEIN_LIST_PATH        ;
extern const char* PROTEIN_LIST_FOLDER  ;
extern const char* PROTEIN_LIST_EXT             ;

extern           char* PEPTIDE_PATH;
extern const char* PEPTIDE_FOLDER;
extern const char* PEPTIDE_EXT;

#include "peptide_tk.h"

#define MAX_SIZE_PROTEIN_LIST   7000
#define MAX_PROT_LENGTH                 5000
#define MAX_NO_REC_SITES                600


enum
{
        SPEPTIDE_MW
};

enum
{
        PROTEOLYSIS_ALL_POSSIBLE, 
        PROTEOLYSIS_COMPLETE
};

typedef struct
{
        int*            GI;
        char**  filename;
} ProteomeIndex;

typedef struct
{
        int*    GI;
        char**  filename;
} ProteinDatabaseIndex;

typedef struct
{
        unsigned flag: 1;
        float mw;
        int   length;
        char* name;
        char* truncated_name;
        char* sequence;
} LProtein;

#include "speptide.h"

/* PROTEASE DEFINITIONS */

#define NO_PROTEASES                    5
#define MAX_PROTEASE_LENGTH             128

/* DIGESTION PARAMETERS */

extern long  SKIP;              /* maximum enzyme digestion skips */
extern long  LCO;               /* maximum lenght of peptide allowed*/


typedef enum {  PROTEASE_NONE, 
                                PROTEASE_TRYPSIN, 
                                PROTEASE_TRYPSIN_CHYMOTRYPSIN, 
                                PROTEASE_CHYMOTRYPSIN,
                                PROTEASE_PEPTIDE_LIBRARY
                        } Enzyme;

extern Enzyme PROTEASE;

int                     Protease_stringToIndex(char* string);
char*           Protease_indexToString(int index);
int                     Protease_set(int enzyme);
int                     isRecSite(char c);


char*           loadProtein(char* szProteinName, char* szFileformat);
void            fillLProteinArrayAtIndexFromFastaFile(LProtein* proteinLibrary, int* index, char* fastafilename);

int                     LProteinArray_loadLibrary(LProtein** proteinLibrary, int* nSizeProteinLibrary, char* szProteinListName);
char*           LProtein_createRestrictionSitesArray(LProtein* prot);
void            LProteinArray_fprint(FILE* fp, LProtein* proteinLibrary, int size);

void            fprintLProteinLibraryNames(FILE* fp, LProtein* proteinLibrary, int size);
LProtein**      extractRibosomalProteins(FILE* fp, LProtein* proteinLibrary, int size, int* setsize);
void            extractProteinsWithName(FILE* fp, LProtein* proteinLibrary, int size, char* name);
void            printProteinSubset(LProtein*** subset);

//Using genbank accession number system load 
int             loadProteomeIndex(ProteomeIndex* PROT_INDEX, long* nProteomeSize);
int             loadProteinDatabaseIndex(ProteinDatabaseIndex* PDBI, long* nProteinDatabaseSize);
long    calculateSizePeptideLibrary(char* szProteomeSubsetName);


long    calculateNumberOfRecSites(char protein[]);
//long  calculateNumberOfRecSites(char protein[], int (bIsRecSite) (char cAminoacid));
long    calculateNumberOfPeptides(char pep[], int options);


void            initializePeptideLibrary                (PeptideLibrary* library);
void            fprintPeptideLibrary                    (FILE* fp, PeptideLibrary library);
long            SPeptideArray_calculateSize             (char protein[], int skipped_sites);

typedef struct
{
        char* restriction_site_array;
        char* boundaries_site_array;
        char** peptide_mark_array;
        LProtein* protein;
        int lines;
        int rem;
        int nrows_per_line;
        int nAminoAcidsPerLine;
} ProteinCoverDisplay;

void ProteinCoverDisplay_fprint(FILE* pOut, ProteinCoverDisplay* pcd);
Range2i* ProteinCoverDisplay_createBoundaryArray(ProteinCoverDisplay* pcd, int* nsize);
int  ProteinCoverDisplay_calcCoverage(ProteinCoverDisplay* pcd);
float* LProtein_getAminoAcidIonCountArray(LProtein* protein, 
                                                                                  char* ion_file, int row_start, int seq_col, int abu_col, int* nPeptides);

#define CHAR_PER_MARK 10

extern ProteomeIndex                    CURR_PROT_INDEX;
extern LProtein*                                CURR_LProteinLibrary;
extern SPeptide*                                CURR_SPeptideLibrary;
extern long                                             nSizePepLibrary;
extern int                                              nSizeProteinLibrary;

#include "proteome/rpeptide.h"
#include "proteome/fastafile.h"
#include "proteome/proteincoverdisplay.h"

#endif

---