bitnuc 0.4.1

A library for efficient nucleotide sequence manipulation using 2-bit and 4-bit encodings
Documentation 
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

use crate::Error;
use std::arch::x86_64::*;

#[allow(unsafe_op_in_unsafe_fn)]
#[inline(always)]
unsafe fn unpack_8_bases(packed: u64, lookup: __m128i) -> __m128i {
    let mut indices = [0u8; 16];

    for (i, v) in indices.iter_mut().enumerate() {
        *v = ((packed >> (i * 2)) & 0b11) as u8;
    }
    let index_vec = _mm_loadu_si128(indices.as_ptr() as *const __m128i);
    _mm_shuffle_epi8(lookup, index_vec)
}

#[allow(unsafe_op_in_unsafe_fn)]
#[inline(always)]
unsafe fn unpack_16_bases(packed: u64, lookup: __m128i) -> __m128i {
    let mut indices = [0u8; 16];
    for (i, v) in indices.iter_mut().enumerate() {
        *v = ((packed >> (i * 2)) & 0b11) as u8;
    }
    let index_vec = _mm_loadu_si128(indices.as_ptr() as *const __m128i);
    _mm_shuffle_epi8(lookup, index_vec)
}

#[allow(unsafe_op_in_unsafe_fn)]
#[inline(always)]
unsafe fn unpack_32_bases(packed: u64, lookup: __m256i) -> __m256i {
    let mut indices = [0u8; 32];
    let bytes = packed.to_le_bytes();

    for (i, &byte) in bytes.iter().enumerate() {
        let base = i * 4;
        indices[base] = byte & 0x03;
        indices[base + 1] = (byte >> 2) & 0x03;
        indices[base + 2] = (byte >> 4) & 0x03;
        indices[base + 3] = (byte >> 6) & 0x03;
    }

    let index_vec = _mm256_loadu_si256(indices.as_ptr() as *const __m256i);
    _mm256_shuffle_epi8(lookup, index_vec)
}

#[allow(unsafe_op_in_unsafe_fn)]
#[inline(always)]
unsafe fn process_remainder(packed: u64, start: usize, end: usize, sequence: &mut Vec<u8>) {
    static LOOKUP: [u8; 4] = [b'A', b'C', b'G', b'T'];
    let count = end - start;
    let old_len = sequence.len();
    sequence.reserve(count);

    let ptr = sequence.as_mut_ptr().add(old_len);
    for i in 0..count {
        let bits = (packed >> ((start + i) * 2)) & 0b11;
        *ptr.add(i) = LOOKUP[bits as usize];
    }
    sequence.set_len(old_len + count);
}

#[allow(unsafe_op_in_unsafe_fn)]
pub unsafe fn from_2bit_simd(
    packed: u64,
    expected_size: usize,
    sequence: &mut Vec<u8>,
) -> Result<(), Error> {
    if expected_size > 32 {
        return Err(Error::InvalidLength(expected_size));
    }

    sequence.reserve(expected_size);

    if expected_size >= 32 {
        // 32 bases at a time
        let lookup = _mm256_setr_epi8(
            b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8,
            b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8,
            b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8,
            b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8,
            b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8,
        );
        let result = unpack_32_bases(packed, lookup);
        let mut temp = [0u8; 32];
        _mm256_storeu_si256(temp.as_mut_ptr() as *mut __m256i, result);
        sequence.extend_from_slice(&temp[..expected_size]);
    } else if expected_size >= 16 {
        // 16 bases at a time
        let lookup = _mm_setr_epi8(
            b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8,
            b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8,
            b'G' as i8, b'T' as i8,
        );
        let simd_chunks = expected_size / 16;
        for chunk in 0..simd_chunks {
            let chunk_data = packed >> (chunk * 32);
            let result = unpack_16_bases(chunk_data, lookup);
            let mut temp = [0u8; 16];
            _mm_storeu_si128(temp.as_mut_ptr() as *mut __m128i, result);
            sequence.extend_from_slice(&temp[..16]);
        }
        let remaining_start = simd_chunks * 16;
        process_remainder(packed, remaining_start, expected_size, sequence);
    } else if expected_size >= 8 {
        // 8 bases at a time
        let lookup = _mm_setr_epi8(
            b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8,
            b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8,
            b'G' as i8, b'T' as i8,
        );
        let simd_chunks = expected_size / 8;
        for chunk in 0..simd_chunks {
            let chunk_data = packed >> (chunk * 16);
            let result = unpack_8_bases(chunk_data, lookup);
            let mut temp = [0u8; 16];
            _mm_storeu_si128(temp.as_mut_ptr() as *mut __m128i, result);
            sequence.extend_from_slice(&temp[..8]);
        }
        let remaining_start = simd_chunks * 8;
        process_remainder(packed, remaining_start, expected_size, sequence);
    } else {
        // Small sequences are handled by the naive implementation
        process_remainder(packed, 0, expected_size, sequence);
    }

    Ok(())
}

#[allow(unsafe_op_in_unsafe_fn)]
#[inline(always)]
pub unsafe fn from_2bit_multi_simd(
    ebuf: &[u64],
    n_bases: usize,
    sequence: &mut Vec<u8>,
) -> Result<(), Error> {
    // Set up SIMD lookup table once for all chunks
    let lookup = _mm256_setr_epi8(
        b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8,
        b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8,
        b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8,
        b'C' as i8, b'G' as i8, b'T' as i8, b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8,
        b'A' as i8, b'C' as i8, b'G' as i8, b'T' as i8,
    );

    // Pre-allocate exact capacity and set length upfront
    let old_len = sequence.len();
    sequence.reserve(n_bases);

    // Get raw pointer to write position
    let mut out_ptr = sequence.as_mut_ptr().add(old_len);

    // Process full 32-base chunks
    let full_chunks = n_bases / 32;

    for &chunk in ebuf.iter().take(full_chunks) {
        let result = unpack_32_bases(chunk, lookup);
        _mm256_storeu_si256(out_ptr as *mut __m256i, result);
        out_ptr = out_ptr.add(32);
    }

    // Handle remaining bases if any
    let remaining_bases = n_bases % 32;
    if remaining_bases > 0 {
        let last_chunk = ebuf[full_chunks];
        let result = unpack_32_bases(last_chunk, lookup);

        // Store to temporary then copy only what we need
        let mut temp = [0u8; 32];
        _mm256_storeu_si256(temp.as_mut_ptr() as *mut __m256i, result);
        std::ptr::copy_nonoverlapping(temp.as_ptr(), out_ptr, remaining_bases);
    }

    // Update the length after all writes are complete
    sequence.set_len(old_len + n_bases);

    Ok(())
}

#[cfg(test)]
mod testing {
    use super::*;
    use crate::as_2bit;

    #[test]
    fn test_from_2bit_simd() {
        let expected = b"ACTGACTGACTGACTGACTG";
        let packed = as_2bit(expected).unwrap();
        let mut observed = Vec::new();
        unsafe {
            from_2bit_simd(packed, 20, &mut observed).unwrap();
        }
        assert_eq!(&observed, expected);
    }

    #[test]
    fn test_various_lengths() {
        for len in 1..=32 {
            let input = b"ACTGACTGACTGACTGACTGACTGACTGACTG";
            let packed = as_2bit(&input[..len]).unwrap();
            let mut observed = Vec::new();
            unsafe {
                from_2bit_simd(packed, len, &mut observed).unwrap();
            }
            assert_eq!(&observed, &input[..len], "Failed at length {}", len);
        }
    }

    #[test]
    fn test_append() {
        let sequence = b"ACTGACTGACTGACTGACTG";
        let packed = as_2bit(sequence).unwrap();
        let mut observed = Vec::new();
        unsafe {
            from_2bit_simd(packed, 10, &mut observed).unwrap();
            from_2bit_simd(packed, 10, &mut observed).unwrap();
        }
        let expected = b"ACTGACTGACACTGACTGAC"; // repeated out of phase
        assert_eq!(&observed, expected);
    }
}