proteome/peptide_tk.h

#ifndef _PEPTIDE_TK_H_
#define _PEPTIDE_TK_H_

#include "massspec/mqaminoacid.h"

enum {NO_ALKYLATION = 1, IODOACETIC_ACID = 2, IODOACETAMIDE = 3};

enum {  MODIFICATION_NONE                                               = 0,
                MODIFICATION_WATER_LOSS                                 = 1,
                MODIFICATION_AMMONIA_LOSS                               = 2,
                MODIFICATION_MET_OXIDATION                              = 3,
                MODIFICATION_CARBAMYLATION                              = 4,
                MODIFICATION_CYS_CARBAMIDOMETHYLATION   = 5,
                MODIFICATION_CYS_CARBOXYMETHYLATION             = 6
};


extern char MODIFICATION_AMMONIA_LOSS_CHAR;
extern char MODIFICATION_MET_OXIDATION_CHAR;
extern char MODIFICATION_CARBAMYLATION_CHAR;
extern char MODIFICATION_CYS_CARBOXYMETHYLATION_CHAR;

void ProteomisToolkit_init();

typedef struct 
{
        char*           name;
        int                     nSize;
        int                     protease;
        int                     protease_skipped_sites;
        int                     nProteinsUsed;

        struct  Peptide* peptides;
} PeptideLibrary;

typedef struct Peptide
{
        unsigned scored :1;     

        /* These fields are assigned by the function getPeptideLibraryFromFile 
           or by the function getPeptideFromString */
        
        char*   protein;

        int             start;
        int             index;
        int             length;
        char*   seq;    
        
        /* These fields are assigned by the function constructPeptideData */
        int             comp[NO_OF_AMINOACIDS];
        float   mw; 
        float   rt;
        float   searchMass;
} Peptide;

#define AASEQ(p) (p.seq)

typedef char* PeptideSequence;

/* Glu and Asp are not deprotonated 
   Arg and Lys are not protonated
   Cys is modified by the reduction/alkylation reaction 160.14 vs 103.14 */

char massToRes(float mass); 
float resToMass(char residue);


int ProteomicsToolkit_setAlkylationAgent(int type);
int ProteomicsToolkit_getAlkylationAgent(char* alkylation_agent);
void ProteomicsToolkit_unsetAlkylationAgent();

float   ProteomicsToolkit_calcMWFromMZ(float mz, int charge);
float   ProteomicsToolkit_calcMZFromMW(float mw, int charge);
void    ProteomicsToolkit_setSequenceModificationSigns();
void    ProteomicsToolkit_oxidizeMethionine();

// PROTEOMICS TOOLKIT 

float   ProteomicsToolkit_calcMW                                        (char protein[], int start_aminoacid, int stop_aminoacid);
float   ProteomicsToolkit_calcHydrophobicity            (char protein[], int start_aminoacid, int stop_aminoacid, int model);
int*    ProteomicsToolkit_calcAminoacidComposition      (char protein[], int start_aminoacid, int stop_aminoacid);
int             ProteomicsToolkit_calcNumberOfCarbons           (char protein[], int start_aminoacid, int stop_aminoacid);

int             ProteomicsToolkit_isPartial(char protein[], int start_aminoacid, int stop_aminoacid);

void    ProteomicsToolkit_getNumberOfChargedResidues(char protein[], int start_aminoacid, int stop_aminoacid,
                                                                                                          int* nD, int* nE, int* nCAMC, 
                                                                                                          int* nH, int* nR, int* nK);

void    ProteomicsToolkit_fprintSubSequence                      (FILE* fp, char protein[], int start, int stop);
void    ProteomicsToolkit_fprintSubSequenceUsingRange(FILE* fp, char protein[], Range2i* range);
void    ProteomicsToolkit_printAminoacidSequence         (char protein[], int start_aminoacid, int stop_aminoacid);
void    ProteomicsToolkit_fprintAminoacidSequence        (FILE* fp, char protein[], int start_aminoacid, int stop_aminoacid);
char*   ProteomicsToolkit_getAminoacidSequence           (char protein[], int start_aminoacid, int stop_aminoacid);
char*   ProteomicsToolkit_getAminoacidSequenceSequestFormat(char protein[], int start_aminoacid, int stop_aminoacid);

// PEPTIDE

Peptide* PeptideArray_create(int nsize);

float   Peptide_calcMW                                  (Peptide* peptide);
float   Peptide_calcHydrophobicity              (Peptide* peptide, int model);
void    Peptide_calcAminoacidComposition(Peptide* peptide);
int             Peptide_calcNumberOfCarbons             (Peptide* peptide);

void    Peptide_printlnAminoacidSequence         (Peptide* peptide); 
void    Peptide_printAminoacidComposition(Peptide* peptide);

void     Peptide_init(Peptide* peptide);
Peptide* Peptide_new (char* sequence);
void     Peptide_createFromString(Peptide* peptide, char* sequence);
void     Peptide_printInfo(Peptide* peptide);

void    printSeparator();



// ***************** PeptideSequence *********************

float   PeptideSequence_calcMW                                  (char* sequence);
float   PeptideSequence_calcHydrophobicity              (char* sequence, int model);
int             PeptideSequence_calcNumberOfCarbons             (char* sequence);
int*    PeptideSequence_calcAminoacidComposition(char* sequence);
int             PeptideSequence_isPartial                               (char* sequence);

Range   PeptideSequence_calcChargeRange                 (char* sequence);

int             PeptideSequence_hasFullyTrypticEnds(char* seq_peptide);
int             PeptideSequence_hasPartiallyTrypticEnds(char* seq_peptide);

void    PeptideSequence_getNumberOfChargedResidues                              (char* sequence, int* nD, int* nE, int *nCAMC, int* nH, int* nR, int* nK);

char*   PeptideSequence_convertFromSequestToSimple                              (char* sequence);
char*   PeptideSequence_convertFromSequestToSimpleWithNoModSigns(char* sequence);
float   PeptideSequence_calcMZ                                                                  (char* sequence, int charge);
char*   PeptideSequence_removeModifications                                             (char* sequence);
int             PeptideSequence_getComposition                                                  (char* sequence, char aa);
int             PeptideSequence_getCompositionSequest(char* seq, char aa);
char*   PeptideSequence_addModification                                                 (char* sequence, char aa, char modif);

#endif

---