src/bayesneti.h

/*****************************************************************************
* This file is provided under the Creative Commons Attribution 3.0 license.
*
* You are free to share, copy, distribute, transmit, or adapt this work
* PROVIDED THAT you attribute the work to the authors listed below.
* For more information, please see the following web page:
* http://creativecommons.org/licenses/by/3.0/
*
* This file is a component of the Sleipnir library for functional genomics,
* authored by:
* Curtis Huttenhower (chuttenh@princeton.edu)
* Mark Schroeder
* Maria D. Chikina
* Olga G. Troyanskaya (ogt@princeton.edu, primary contact)
*
* If you use this library, the included executable tools, or any related
* code in your work, please cite the following publication:
* Curtis Huttenhower, Mark Schroeder, Maria D. Chikina, and
* Olga G. Troyanskaya.
* "The Sleipnir library for computational functional genomics"
*****************************************************************************/
#ifndef BAYESNETI_H
#define BAYESNETI_H

#ifndef NO_SMILE
#include <smile.h>
#include <syscoord.h>
#endif // NO_SMILE

#include "dataset.h"
//#include "trie.h"

namespace pnl {

class CBNet;

}

namespace Sleipnir {

class CBayesNetPNL;
class CBayesNetSmile;
class CPCLPair;

class CBayesNetImpl {
protected:
    static const size_t c_iMinimum  = 25;
    static const char   c_cMissing  = '_';
    static const char   c_cBase     = 'A';
    static const char   c_szFR[];
    static const char   c_szZero[];

    typedef std::vector<std::vector<float> >    TVecVecD;
    typedef std::map<std::string, size_t>       TMapData;
//  typedef CTrie<size_t>                       TTrieData;

    CBayesNetImpl( bool );

    static void EncodeData( const IDataset*, TMapData& );
    static std::string EncodeDatum( const IDataset*, size_t, size_t );
    static std::string EncodeDatum( const CPCLPair&, size_t, const std::vector<size_t>& );
    static void DecodeDatum( const std::string&, std::vector<size_t>& );
    static bool IsAnswer( const std::string& );

    bool    m_fGroup;
};

#ifndef NO_SMILE

class CBayesNetSmileImpl : protected CBayesNetImpl {
protected:
    friend class CBayesNetFN;

    static const char   c_szGaussian[];

    static bool IsGaussian( const DSL_network& );
    static bool IsNaive( const DSL_network& );
    static float ELRDot( const TVecVecD&, const TVecVecD& );
    static float ELRAvoidZero( float );
    static void ELRComputeDirection( float, const TVecVecD&, TVecVecD& );
    static bool GetCPT( DSL_node*, CDataMatrix& );

    CBayesNetSmileImpl( bool );

    bool ConvertGraph( CBayesNetPNL& ) const;
    bool ConvertCPTs( CBayesNetPNL& ) const;
    void LearnExpected( DSL_node*, DSL_Dmatrix*, size_t = 1 );
    bool Evaluate( const IDataset*, CDat*, TVecVecD*, bool ) const;
    bool FillCPTs( const IDataset*, size_t, size_t, bool, bool );
    bool FillCPTs( const std::vector<bool>&, const std::string&, bool, bool, bool = false );
    bool FillCPTs( const std::vector<bool>&, const std::vector<unsigned char>&, bool, bool, bool = false );
    bool LearnGrouped( const IDataset*, size_t, bool );
    bool LearnUngrouped( const IDataset*, size_t, bool );
    bool LearnNaive( const IDataset*, bool );
    bool LearnELR( const IDataset*, size_t, bool );
    size_t ELRCountParameters( ) const;
    void ELRCopyParameters( TVecVecD& );
    void ELRComputeGradient( const std::vector<bool>&, const TMapData&, bool, TVecVecD& );
    void ELRUpdateGradient( float, TVecVecD& );
    void ELRNormalizeDirection( TVecVecD& ) const;
    float ELRLineSearch( const std::vector<bool>&, const TMapData&, const TVecVecD&, const TVecVecD&, TVecVecD&,
        float&, float&, bool );
    float ELREvalFunction( const std::vector<bool>&, const TMapData&, float, const TVecVecD&, const TVecVecD&,
        TVecVecD&, bool );
    void ELRBracket( const std::vector<bool>&, const TMapData&, const TVecVecD&, const TVecVecD&, TVecVecD&,
        float&, float&, float&, float&, float&, float&, bool );
    float ELRConditionalLikelihood( const std::vector<bool>&, const TMapData&, bool );
    float ELRBrent( const std::vector<bool>&, const TMapData&, const TVecVecD&, const TVecVecD&, TVecVecD&,
        float&, float&, float, float, float, float, bool );

    bool IsNaive( ) const {

        return ( m_fSmileNet ? CBayesNetSmileImpl::IsNaive( m_SmileNet ) : false ); }

    bool IsContinuous( ) const {

        return ( m_fSmileNet ? IsGaussian( m_SmileNet ) : false ); }

    bool                    m_fSmileNet;
    DSL_network             m_SmileNet;
    const CBayesNetSmile*   m_pDefaults;
};

#endif // NO_SMILE

class CBayesNetPNLImpl : protected CBayesNetImpl {
protected:
    friend class CBayesNetSmileImpl;

    static const char   c_szBN[];

    CBayesNetPNLImpl( bool );
    ~CBayesNetPNLImpl( );

    bool Evaluate( const IDataset*, CDat*, TVecVecD*, bool ) const;
    bool IsContinuous( ) const;

    pnl::CBNet* m_pPNLNet;
};

}

#include "bayesnetfni.h"

#endif // BAYESNETI_H