seqan3::detail::simd_affine_gap_policy< alignment_algorithm_t, score_t, align_local_t > Class Template Reference

The CRTP-policy that computes a batch of cells in the alignment matrix using simd instructions. More...

#include <seqan3/alignment/pairwise/policy/simd_affine_gap_policy.hpp>

Private Types
using	alignment_state_t = alignment_algorithm_state< score_t >
	The type of state of the alignment algorithm for affine gaps.

Private Member Functions
template<typename cell_t >
constexpr void	compute_cell (cell_t &&current_cell, alignment_algorithm_state< score_t > &state, score_t const score) const noexcept
	Computes the score of the current simd cell.
constexpr score_t	convert_to_simd (trace_directions const direction) const noexcept
	Converts a trace direction into a simd vector.
template<typename alignment_configuration_t >
constexpr void	initialise_alignment_state (alignment_configuration_t const &config) noexcept
	Initialise the alignment state for affine gap computation.
Constructors, destructor and assignment
constexpr	simd_affine_gap_policy () noexcept=default
	Defaulted.
constexpr	simd_affine_gap_policy (simd_affine_gap_policy const &) noexcept=default
	Defaulted.
constexpr	simd_affine_gap_policy (simd_affine_gap_policy &&) noexcept=default
	Defaulted.
constexpr simd_affine_gap_policy &	operator= (simd_affine_gap_policy const &) noexcept=default
	Defaulted.
constexpr simd_affine_gap_policy &	operator= (simd_affine_gap_policy &&) noexcept=default
	Defaulted.
	~simd_affine_gap_policy () noexcept=default
	Defaulted.
template<typename configuration_t >
	simd_affine_gap_policy (configuration_t const &)
	Initialise the policy.

Private Attributes
friend	alignment_algorithm_t
	Befriends the derived class to grant it access to the private members.
alignment_state_t	alignment_state {}
	The internal alignment state tracking the current alignment optimum.

Detailed Description

template<typename alignment_algorithm_t, simd_concept score_t, typename align_local_t = std::false_type>
class seqan3::detail::simd_affine_gap_policy< alignment_algorithm_t, score_t, align_local_t >

The CRTP-policy that computes a batch of cells in the alignment matrix using simd instructions.

Template Parameters

alignment_algorithm_t	The derived type (seqan3::detail::alignment_algorithm) to be augmented with this CRTP-policy.
score_t	The score type of the dynamic programming matrix; must model seqan3::simd::simd_concept.
align_local_t	A std::bool_constant to switch between local and global alignment.

This CRTP-policy implements the recursion for the alignment algorithm with affine gaps using an inter-sequence vectorisation scheme. See Rahn, R, et al. Generic accelerated sequence alignment in SeqAn using vectorization and multi-threading. Bioinformatics 34.20 (2018): 3437-3445. for more information.

Remarks

The template parameters of this CRTP-policy are selected in the seqan3::detail::alignment_configurator::select_gap_policy when selecting the alignment for the given configuration.

Member Function Documentation

compute_cell()

template<typename alignment_algorithm_t , simd_concept score_t, typename align_local_t = std::false_type>

template<typename cell_t >

constexpr void seqan3::detail::simd_affine_gap_policy< alignment_algorithm_t, score_t, align_local_t >::compute_cell ( cell_t &&  current_cell, alignment_algorithm_state< score_t > &  state, score_t const  score  ) const

inlineconstexprprivatenoexcept

Computes the score of the current simd cell.

Template Parameters

cell_t

The type of the current cell [for detailed information on the type see below].

Parameters

[in,out]	current_cell	The current cell in the dynamic programming matrix.
[in,out]	state	The state storing hot helper variables.
[in]	score	The score of comparing the respective letters of the first and the second sequence.

cell_t is the result type of dereferencing the zipped iterator over the respective alignment score matrix and the alignment trace matrix used inside of the seqan3::detail::alignment_matrix_policy. The first parameter stored in the zipped tuple is the seqan3::detail::alignment_score_matrix_proxy and the second value is the seqan3::detail::alignment_trace_matrix_proxy.

In order to compute the maximum for two simd vectors, gcc implements the ternary operator such that std::max(a, b) can be implemented as (a > b) ? a : b, where (a > b) returns a mask vector. This implements the compare-and-blend approach for simd vector types.

convert_to_simd()

template<typename alignment_algorithm_t , simd_concept score_t, typename align_local_t = std::false_type>

constexpr score_t seqan3::detail::simd_affine_gap_policy< alignment_algorithm_t, score_t, align_local_t >::convert_to_simd ( trace_directions const  direction ) const

inlineconstexprprivatenoexcept

Converts a trace direction into a simd vector.

Parameters

[in]

direction

The trace direction to convert to a simd vector.

initialise_alignment_state()

template<typename alignment_algorithm_t , simd_concept score_t, typename align_local_t = std::false_type>

template<typename alignment_configuration_t >

constexpr void seqan3::detail::simd_affine_gap_policy< alignment_algorithm_t, score_t, align_local_t >::initialise_alignment_state ( alignment_configuration_t const &  config )

inlineconstexprprivatenoexcept

Initialise the alignment state for affine gap computation.

Template Parameters

alignment_configuration_t

The type of alignment configuration.

Parameters

[in]

config

The alignment configuration.

Gets the stored gap scheme from the configuration or uses a default gap scheme if not set and initialises the alignment algorithm state with the respective gap extension and gap open costs. If the gap scheme was not specified by the user the following defaults are used:

• -1 for the gap extension score, and
• -10 for the gap open score.

---

The documentation for this class was generated from the following file:

• simd_affine_gap_policy.hpp