1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227

use crate::structs::dp_matrix::DpMatrix;
use crate::structs::trace::constants::{
    TRACE_B, TRACE_C, TRACE_D, TRACE_E, TRACE_I, TRACE_J, TRACE_M, TRACE_N, TRACE_S, TRACE_T,
};
use crate::structs::{Profile, Trace};

pub fn traceback_bounded(
    profile: &Profile,
    posterior_matrix: &impl DpMatrix,
    optimal_matrix: &impl DpMatrix,
    trace: &mut Trace,
    target_end: usize,
) {
    let mut target_idx = target_end;
    let mut profile_idx = 0;

    let mut current_state_posterior_probability: f32;
    // we trace back starting from the last C state
    let mut previous_state: usize = TRACE_C;
    let mut current_state: usize;

    trace.append_with_posterior_probability(TRACE_T, target_idx, profile_idx, 0.0);
    trace.append_with_posterior_probability(TRACE_C, target_idx, profile_idx, 0.0);

    while previous_state != TRACE_S {
        current_state = match previous_state {
            TRACE_C => {
                let c_to_c_path = profile.special_transition_score_delta(
                    Profile::SPECIAL_C_IDX,
                    Profile::SPECIAL_LOOP_IDX,
                ) * (optimal_matrix
                    .get_special(target_idx - 1, Profile::SPECIAL_C_IDX)
                    + posterior_matrix.get_special(target_idx, Profile::SPECIAL_C_IDX));

                let c_to_e_path = profile
                    .transition_score_delta(Profile::SPECIAL_E_IDX, Profile::SPECIAL_MOVE_IDX)
                    * optimal_matrix.get_special(target_idx, Profile::SPECIAL_E_IDX);

                if c_to_c_path > c_to_e_path {
                    TRACE_C
                } else {
                    TRACE_E
                }
            }
            TRACE_E => {
                let mut max_score = -f32::INFINITY;
                let mut state_of_max_score = 0;
                let mut profile_idx_of_max_score = 0;

                for profile_idx in 1..=profile.length {
                    if optimal_matrix.get_match(target_idx, profile_idx) >= max_score {
                        max_score = optimal_matrix.get_match(target_idx, profile_idx);
                        state_of_max_score = TRACE_M;
                        profile_idx_of_max_score = profile_idx;
                    }
                    if optimal_matrix.get_delete(target_idx, profile_idx) > max_score {
                        max_score = optimal_matrix.get_delete(target_idx, profile_idx);
                        state_of_max_score = TRACE_D;
                        profile_idx_of_max_score = profile_idx;
                    }
                }
                profile_idx = profile_idx_of_max_score;
                state_of_max_score
            }
            TRACE_M => {
                let possible_states: [usize; 4] = [TRACE_M, TRACE_I, TRACE_D, TRACE_B];
                let possible_paths: [f32; 4] = [
                    profile.transition_score_delta(Profile::MATCH_TO_MATCH_IDX, profile_idx - 1)
                        * optimal_matrix.get_match(target_idx - 1, profile_idx - 1),
                    profile.transition_score_delta(Profile::INSERT_TO_MATCH_IDX, profile_idx - 1)
                        * optimal_matrix.get_insert(target_idx - 1, profile_idx - 1),
                    profile.transition_score_delta(Profile::DELETE_TO_MATCH_IDX, profile_idx - 1)
                        * optimal_matrix.get_delete(target_idx - 1, profile_idx - 1),
                    profile.transition_score_delta(Profile::BEGIN_TO_MATCH_IDX, profile_idx - 1)
                        * optimal_matrix.get_special(target_idx - 1, Profile::SPECIAL_B_IDX),
                ];

                let mut argmax: usize = 0;
                for i in 1..4 {
                    if possible_paths[i] > possible_paths[argmax] {
                        argmax = i;
                    }
                }

                // a match means we have moved forward in the both the profile and the target
                profile_idx -= 1;
                target_idx -= 1;

                possible_states[argmax]
            }
            TRACE_I => {
                let match_to_insert_path = profile
                    .transition_score_delta(Profile::MATCH_TO_INSERT_IDX, profile_idx)
                    * optimal_matrix.get_match(target_idx - 1, profile_idx);
                let insert_to_insert_path: f32 = profile
                    .transition_score_delta(Profile::INSERT_TO_INSERT_IDX, profile_idx)
                    * optimal_matrix.get_insert(target_idx - 1, profile_idx);

                // an insert means we moved forward only in the profile
                target_idx -= 1;

                if match_to_insert_path >= insert_to_insert_path {
                    TRACE_M
                } else {
                    TRACE_I
                }
            }
            TRACE_D => {
                let match_to_delete_path = profile
                    .transition_score_delta(Profile::MATCH_TO_DELETE_IDX, profile_idx - 1)
                    * optimal_matrix.get_match(target_idx, profile_idx - 1);
                let delete_to_delete_path = profile
                    .transition_score_delta(Profile::DELETE_TO_DELETE_IDX, profile_idx - 1)
                    * optimal_matrix.get_delete(target_idx, profile_idx - 1);

                // a delete means we moved forward only in the profile
                profile_idx -= 1;

                if match_to_delete_path >= delete_to_delete_path {
                    TRACE_M
                } else {
                    TRACE_D
                }
            }
            TRACE_B => {
                let n_to_b_path = profile.special_transition_score_delta(
                    Profile::SPECIAL_N_IDX,
                    Profile::SPECIAL_MOVE_IDX,
                ) * optimal_matrix.get_match(target_idx, Profile::SPECIAL_N_IDX);
                let j_to_b_path = profile
                    .transition_score_delta(Profile::SPECIAL_J_IDX, Profile::SPECIAL_MOVE_IDX)
                    * optimal_matrix.get_special(target_idx, Profile::SPECIAL_J_IDX);

                if n_to_b_path >= j_to_b_path {
                    TRACE_N
                } else {
                    TRACE_J
                }
            }
            TRACE_N => {
                if target_idx == 0 {
                    TRACE_S
                } else {
                    TRACE_N
                }
            }
            TRACE_J => {
                let j_to_j_path = profile.special_transition_score_delta(
                    Profile::SPECIAL_J_IDX,
                    Profile::SPECIAL_LOOP_IDX,
                ) * (optimal_matrix
                    .get_special(target_idx - 1, Profile::SPECIAL_J_IDX)
                    + posterior_matrix.get_special(target_idx, Profile::SPECIAL_J_IDX));
                let e_to_j_path = profile
                    .transition_score_delta(Profile::SPECIAL_E_IDX, Profile::SPECIAL_LOOP_IDX)
                    * optimal_matrix.get_special(target_idx, Profile::SPECIAL_E_IDX);

                if j_to_j_path > e_to_j_path {
                    TRACE_J
                } else {
                    TRACE_E
                }
            }
            _ => {
                panic!("bad state in traceback")
            }
        };

        current_state_posterior_probability = get_posterior_probability(
            posterior_matrix,
            current_state,
            previous_state,
            profile_idx,
            target_idx,
        );

        trace.append_with_posterior_probability(
            current_state,
            target_idx,
            profile_idx,
            current_state_posterior_probability,
        );

        if (current_state == TRACE_N || current_state == TRACE_J || current_state == TRACE_C)
            && current_state == previous_state
        {
            target_idx -= 1;
        }
        previous_state = current_state;
    }
    trace.reverse();
}

pub fn get_posterior_probability(
    optimal_matrix: &impl DpMatrix,
    current_state: usize,
    previous_state: usize,
    profile_idx: usize,
    target_idx: usize,
) -> f32 {
    match current_state {
        TRACE_M => optimal_matrix.get_match(target_idx, profile_idx),
        TRACE_I => optimal_matrix.get_insert(target_idx, profile_idx),
        TRACE_N => {
            if current_state == previous_state {
                optimal_matrix.get_special(target_idx, Profile::SPECIAL_N_IDX)
            } else {
                0.0
            }
        }
        TRACE_C => {
            if current_state == previous_state {
                optimal_matrix.get_special(target_idx, Profile::SPECIAL_C_IDX)
            } else {
                0.0
            }
        }
        TRACE_J => {
            if current_state == previous_state {
                optimal_matrix.get_special(target_idx, Profile::SPECIAL_J_IDX)
            } else {
                0.0
            }
        }
        _ => 0.0,
    }
}