common.h

Go to the documentation of this file.

 
#ifndef COMMON_H
#define COMMON_H
 
#include <vector>
#include <string>
 
const std::vector<std::string> VALID_FORMATS = {"fasta", "afa", "fas", "fst", "fsa", "a2m", "a3m", "sto", "clustal", "aln", "pfam"};
 
inline const std::vector<std::string> FASTA_FORMATS = {"fasta", "afa", "fas", "fst", "fsa"};
 
const std::string STANDARD_AMINO_ACIDS = "ACDEFGHIKLMNPQRSTVWY";
const std::string NON_STANDARD_AMINO_ACIDS = "XOUBJZ";
const std::string STANDARD_RNA_NUCLEOTIDES = "AUCG";
const std::string STANDARD_DNA_NUCLEOTIDES = "ATCG";
const std::string NON_STANDARD_NUCLEOTIDES = "N";
const std::string GAP = "-.";
 
// Structure to represent a sequence
struct Sequence
{
    std::string id;       // Sequence identifier
    std::string sequence; // Sequence data
    std::string remarks;  // Remarks associated with the sequence
};
 
// Enumeration for NEFF normalization options
enum Normalization
{
    Sqrt_L = 0, // Square root of the length
    L,          // Length
    None        // No normalization
};
 
// Enumeration for MSA alphabet
enum Alphabet
{
    protein = 0,
    RNA,
    DNA
};
 
// Enumeration for options we have for handling non-standard residues in neff computation
enum NonStandardHandler
{
    AsStandard = 0, // treat the the same as standard residues, default in deepMSA symmetric
    ConsiderGapInCutoff, // consider them as gaps when computing cutoff of sequence (only used in asymmetryc version, default in deepMSA assymetric)
    ConsiderGap // consider them as gaps in computing similarity cutoff and checking position match/mismatch, default in Gremlin
};
 
std::string getFormat(std::string file, std::string name);
 
void keepNonGapPositionsOfQuerySequence(std::vector<Sequence>& sequences);
 
std::string getAllowedLetters(Alphabet alphabet);
 
std::string getStandardLetters(Alphabet alphabet);
 
std::string getNonStandardLetters(Alphabet alphabet);
#endif