void
profileToConsensus (
ProfileType const & profile,
Sequence<ResidueType> & sequence
)
{
uint32_t profile_length = profile.length();
sequence.reinitialize( profile_length );
for( uint32_t pos_i = 0; pos_i < profile_length; pos_i++ ) {
sequence[ pos_i ] =
profile[ pos_i ][ Emission::Match ].maximumValueType();
}
// That's it.
return;
} // profileToConsensus( ProfileType const &, Sequence & )
int
convert_to_galosh_profile ( P7_HMM * hmm, ProfileType & profile )
{
typedef typename galosh::profile_traits<ProfileType>::ResidueType ResidueType;
int status;
uint32_t pos_i; // Position in profile. Corresponds to one less than match state pos in HMM.
uint32_t res_i;
ESL_DSQ hmmer_digitized_residue;
/* How many match states in the HMM? */
if( hmm->M == 0 ) { status = eslENORESULT; goto ERROR; }
profile.reinitialize( static_cast<uint32_t>( hmm->M ) );
profile.zero();
// NOTE that HMMER3 has a slightly different model, starting in
// Begin rather than in preAlign, and with 3 legal transitions out
// of Begin (one of these is to PreAlign). The galosh profile model
// begins in preAlign and transitions to Begin, and from there to
// either Match or Delete. One implication is that galosh profiles
// enforce t[ 0 ][ p7H_MI ] to be the same as t[ 0 ][ p7H_II ], but
// HMMER3 does not. Another way to say this is that H3 uses affine
// pre-aligns, and prohibits pre-align -to- delete transitions,
// whereas galosh / profillic uses non-affine pre-aligns and allows
// pre-align->delete.
// fromPreAlign
profile[ galosh::Transition::fromPreAlign ][ galosh::TransitionFromPreAlign::toPreAlign ] =
hmm->t[ 0 ][ p7H_II ];
profile[ galosh::Transition::fromPreAlign ][ galosh::TransitionFromPreAlign::toBegin ] =
hmm->t[ 0 ][ p7H_IM ];
for( res_i = 0; res_i < seqan::ValueSize<ResidueType>::VALUE; res_i++ ) {
hmmer_digitized_residue =
esl_abc_DigitizeSymbol( hmm->abc, static_cast<char>( ResidueType( res_i ) ) );
// See below where it says "TODO/NOTE"..
profile[ galosh::Emission::PreAlignInsertion ][ res_i ] =
hmm->ins[ 0 ][ hmmer_digitized_residue ];
}
// fromBegin
profile[ galosh::Transition::fromBegin ][ galosh::TransitionFromBegin::toMatch ] =
( hmm->t[ 0 ][ p7H_MM ] / ( 1.0 - hmm->t[ 0 ][ p7H_MI ] ) );
profile[ galosh::Transition::fromBegin ][ galosh::TransitionFromBegin::toDeletion ] =
( 1.0 - profile[ galosh::Transition::fromBegin ][ galosh::TransitionFromBegin::toMatch ] );
for( pos_i = 0; pos_i < profile.length(); pos_i++ ) {
// if( be_verbose ) {
// cout << '.';
// cout.flush();
// }
// TODO: If this is too slow, memoize the ResidueType( res_i )s.
for( res_i = 0; res_i < seqan::ValueSize<ResidueType>::VALUE; res_i++ ) {
hmmer_digitized_residue =
esl_abc_DigitizeSymbol( hmm->abc, static_cast<char>( ResidueType( res_i ) ) );
profile[ pos_i ][ galosh::Emission::Match ][ res_i ] =
hmm->mat[ pos_i + 1 ][ hmmer_digitized_residue ];
if( pos_i == ( profile.length() - 1 ) ) {
// Use post-align insertions
profile[ galosh::Emission::PostAlignInsertion ][ res_i ] =
hmm->ins[ pos_i + 1 ][ hmmer_digitized_residue ];
} else { // if this is the last position (use post-align insertions) .. else ..
profile[ galosh::Emission::Insertion ][ res_i ] +=
hmm->ins[ pos_i + 1 ][ hmmer_digitized_residue ];
} // End if this is the last position (use post-align insertions) .. else ..
} // End foreach res_i
if( pos_i == ( profile.length() - 1 ) ) {
// Use post-align insertions
profile[ galosh::Transition::fromPostAlign ][ galosh::TransitionFromPostAlign::toTerminal ] =
hmm->t[ pos_i + 1 ][ p7H_IM ];
profile[ galosh::Transition::fromPostAlign ][ galosh::TransitionFromPostAlign::toPostAlign ] =
( 1.0 - profile[ galosh::Transition::fromPostAlign ][ galosh::TransitionFromPostAlign::toTerminal ] );
} else { // if this is the last position (use post-align insertions) .. else ..
profile[ galosh::Transition::fromMatch ][ galosh::TransitionFromMatch::toMatch ] +=
hmm->t[ pos_i + 1 ][ p7H_MM ];
profile[ galosh::Transition::fromMatch ][ galosh::TransitionFromMatch::toInsertion ] +=
hmm->t[ pos_i + 1 ][ p7H_MI ];
profile[ galosh::Transition::fromMatch ][ galosh::TransitionFromMatch::toDeletion ] +=
hmm->t[ pos_i + 1 ][ p7H_MD ];
profile[ galosh::Transition::fromInsertion ][ galosh::TransitionFromInsertion::toMatch ] +=
hmm->t[ pos_i + 1 ][ p7H_IM ];
profile[ galosh::Transition::fromInsertion ][ galosh::TransitionFromInsertion::toInsertion ] +=
hmm->t[ pos_i + 1 ][ p7H_II ];
profile[ galosh::Transition::fromDeletion ][ galosh::TransitionFromDeletion::toMatch ] +=
hmm->t[ pos_i + 1 ][ p7H_DM ];
profile[ galosh::Transition::fromDeletion ][ galosh::TransitionFromDeletion::toDeletion ] +=
hmm->t[ pos_i + 1 ][ p7H_DD ];
} // End if this is the last position (use post-align insertions) .. else ..
} // End foreach pos_i
// Normalize with 0 as the minimum value we'll allow. Note that in
// profillic and profuse, it's generally 1E-5, so when the profile
// is read in by those programs, it might be slightly altered.
profile.normalize( 0 );
return eslOK;
ERROR:
return status;
} // convert_to_galosh_profile (..)
