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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978
/*
* Copyright (c) 2020-2024 COMBINE-lab.
*
* This file is part of libradicl
* (see https://www.github.com/COMBINE-lab/libradicl).
*
* License: 3-clause BSD, see https://opensource.org/licenses/BSD-3-Clause
*/
//! `libradicl` is a crate for reading (and writing) RAD (Reduced Alignment Data) format
//! files. The RAD format is a binary format designed to encode alignment information
//! about sequencing reads and how they map to a set of targets (a genome, metagenome,
//! transcriptome, etc.). The format is "reduced" because it is allowed to contain sparser
//! information than e.g. a [SAM](https://samtools.github.io/hts-specs/) format file.
//!
//!
//! While the eventual goal of this crate is to provide a generic API to read and write RAD
//! files that may be designed for any purpose, it is driven mostly by our (the [COMBINE-lab's](https://combine-lab.github.io/))
//! needs within the tools we produce that use the RAD format (e.g. [`alevin-fry`](https://github.com/COMBINE-lab/alevin-fry) and
//! [`piscem-infer`](https://github.com/COMBINE-lab/alevin-fry)). Thus, features are generally
//! developed and added in the order that is most urgent to the development of these tools.
//! However, we welcome external contributions via pull requests, and are happy to discuss
//! your potential use cases for the RAD format, and how they might be supported.
//!
//! This crate is broken into several components that cover the various parts of RAD files
//! including the type tag system, the header, and the main data chunks. The names of
//! each module are fairly self-explanatory.
//!
// scroll now, explore nom later
use crate as libradicl;
use self::libradicl::rad_types::RadIntId;
use self::libradicl::record::AlevinFryReadRecord;
use self::libradicl::record::AtacSeqReadRecord;
use self::libradicl::schema::TempCellInfo;
#[allow(unused_imports)]
use ahash::{AHasher, RandomState};
use bio_types::strand::*;
use scroll::Pread;
use serde::{Deserialize, Serialize};
use std::collections::HashMap;
use std::fs::File;
use std::io::{BufReader, Cursor, Read, Seek, SeekFrom};
use std::io::{BufWriter, Write};
use std::sync::atomic::{AtomicU32, AtomicU64, Ordering};
use std::sync::{Arc, Mutex};
use std::vec::Vec;
pub mod chunk;
pub mod constants;
pub mod exit_codes;
pub mod header;
pub mod io;
pub mod rad_types;
pub mod readers;
pub mod record;
pub mod schema;
pub mod utils;
#[macro_use]
mod macros;
// Name of the program, to be used in diagnostic messages.
static LIB_NAME: &str = "libradicl";
pub fn lib_name() -> &'static str {
LIB_NAME
}
#[derive(Serialize, Deserialize, Debug)]
#[allow(dead_code)]
pub struct BarcodeLookupMap {
pub barcodes: Vec<u64>,
//pub counts: Vec<usize>,
offsets: Vec<usize>,
bclen: u32,
prefix_len: u32,
suffix_len: u32,
}
impl BarcodeLookupMap {
pub fn new(mut kv: Vec<u64>, bclen: u32) -> BarcodeLookupMap {
let prefix_len = ((bclen + 1) / 2) as u64;
let suffix_len = bclen - prefix_len as u32;
let _prefix_bits = 2 * prefix_len;
let suffix_bits = 2 * suffix_len;
kv.sort_unstable();
let pref_mask = ((4usize.pow(prefix_len as u32) - 1) as u64) << (suffix_bits);
let mut offsets = vec![0; 4usize.pow(prefix_len as u32) + 1];
let mut prev_ind = 0xFFFF;
for (n, &v) in kv.iter().enumerate() {
let ind = ((v & pref_mask) >> (suffix_bits)) as usize;
if ind != prev_ind {
for item in offsets.iter_mut().take(ind).skip(prev_ind + 1) {
*item = n;
}
offsets[ind] = n;
prev_ind = ind;
}
}
for item in offsets.iter_mut().skip(prev_ind + 1) {
*item = kv.len();
}
//let nbc = kv.len();
BarcodeLookupMap {
barcodes: kv,
//counts: vec![0usize; nbc],
offsets,
bclen,
prefix_len: prefix_len as u32,
suffix_len,
}
}
#[allow(dead_code)]
pub fn barcode_for_idx(&self, idx: usize) -> u64 {
self.barcodes[idx]
}
pub fn find_exact(&self, query: u64) -> Option<usize> {
let mut ret: Option<usize> = None;
// extract the prefix we will use to search
let suffix_bits = 2 * self.suffix_len;
let query_pref = query >> suffix_bits;
// the range of entries having query_pref as their prefix
let qrange = std::ops::Range {
start: self.offsets[query_pref as usize],
end: self.offsets[(query_pref + 1) as usize],
};
let qs = qrange.start;
// if we can, then we return the found barcode and that there was 1 best hit
if let Ok(res) = self.barcodes[qrange].binary_search(&query) {
ret = Some(qs + res);
}
ret
}
/// The find function searches for the barcode `query` in the
/// BarcodeLookupMap. It returns a tuple `(Option<usize>, usize)` where
/// the first element is either Some(usize) or None. If
/// Some(usize) is returned, this is the *index* of a matching/neighboring barcode
/// if None is returned, then no match was found. The second element is either
/// 0, 1 or 2. If 0, no match was found; if 1 a unique match was found, if 2
/// then 2 or more equally good matches were found.
///
/// The parameter `try_exact` controls whether a an exact search is performed
/// or not. If this parameter is true, an exact search is performed before
/// a neighbor search. Otherwise, the exact search is skipped.
pub fn find_neighbors(&self, query: u64, try_exact: bool) -> (Option<usize>, usize) {
let mut ret: Option<usize> = None;
// extract the prefix we will use to search
let pref_bits = 2 * self.prefix_len;
let suffix_bits = 2 * self.suffix_len;
let mut query_pref = query >> suffix_bits;
let mut num_neighbors = 0usize;
// the range of entries having query_pref as their prefix
let qrange = std::ops::Range {
start: self.offsets[query_pref as usize],
end: self.offsets[(query_pref + 1) as usize],
};
let qs = qrange.start;
if try_exact {
// first, we try to find exactly.
// if we can, then we return the found barcode and that there was 1 best hit
if let Ok(res) = self.barcodes[qrange.clone()].binary_search(&query) {
ret = Some(qs + res);
num_neighbors += 1;
return (ret, num_neighbors);
}
}
// othwerwise, we fall back to the 1 mismatch search
// NOTE: We stop here as soon as we find at most 2 neighbors
// for the query. Thus, we only distinguish between the
// the cases where the query has 1 neighbor, or 2 or more neighbors.
// if we match the prefix exactly, we will look for possible matches
// that are 1 mismatch off in the suffix.
if !(std::ops::Range::<usize>::is_empty(&qrange)) {
// the initial offset of suffixes for this prefix
let qs = qrange.start;
// for each position in the suffix
for i in (0..suffix_bits).step_by(2) {
let bit_mask = 3 << (i);
// for each nucleotide
for nmod in 1..4 {
let nucl = 0x3 & ((query >> i) + nmod);
let nquery = (query & (!bit_mask)) | (nucl << i);
if let Ok(res) = self.barcodes[qrange.clone()].binary_search(&nquery) {
ret = Some(qs + res);
num_neighbors += 1;
if num_neighbors >= 2 {
return (ret, num_neighbors);
}
}
}
}
}
{
// if we get here we've had either 0 or 1 matches holding the prefix fixed
// so we will now hold the suffix fixed and consider possible mutations of the prefix.
// for each position in the prefix
for i in (suffix_bits..(suffix_bits + pref_bits)).step_by(2) {
let bit_mask = 3 << i;
// for each nucleotide
for nmod in 1..4 {
let nucl = 0x3 & ((query >> i) + nmod);
let nquery = (query & (!bit_mask)) | (nucl << i);
query_pref = nquery >> suffix_bits;
let qrange = std::ops::Range {
start: self.offsets[query_pref as usize],
end: self.offsets[(query_pref + 1) as usize],
};
let qs = qrange.start;
if let Ok(res) = self.barcodes[qrange].binary_search(&nquery) {
ret = Some(qs + res);
num_neighbors += 1;
if num_neighbors >= 2 {
return (ret, num_neighbors);
}
}
}
}
}
(ret, num_neighbors)
}
}
#[derive(Debug, Clone)]
pub struct GlobalEqCellList {
cell_ids: Vec<usize>,
count: u32,
}
impl GlobalEqCellList {
pub fn from_umi_and_count(bc_mer: usize, count: u32) -> GlobalEqCellList {
let mut cc = GlobalEqCellList {
cell_ids: Vec::new(),
count: 0,
};
cc.cell_ids.push(bc_mer);
cc.count += count;
cc
}
pub fn add_element(&mut self, bc_mer: usize, count: u32) {
self.cell_ids.push(bc_mer);
self.count += count;
}
}
/*
#[inline]
pub fn dump_chunk(v: &mut CorrectedCbChunk, owriter: &Mutex<BufWriter<File>>) {
v.data.set_position(0);
let nbytes = (v.data.get_ref().len()) as u32;
let nrec = v.nrec;
v.data.write_all(&nbytes.to_le_bytes()).unwrap();
v.data.write_all(&nrec.to_le_bytes()).unwrap();
owriter.lock().unwrap().write_all(v.data.get_ref()).unwrap();
}
*/
/// Given a [BufReader]`<T>` from which to read a set of records that
/// should reside in the same collated bucket, this function will
/// collate the records by cell barcode, filling them into a chunk of
/// memory exactly as they will reside on disk. If `compress` is true
/// the collated chunk will be compressed, and then the result will be
/// written to the output guarded by `owriter`.
pub fn collate_temporary_bucket_twopass<T: Read + Seek, U: Write>(
reader: &mut BufReader<T>,
bct: &RadIntId,
umit: &RadIntId,
nrec: u32,
owriter: &Mutex<U>,
compress: bool,
cb_byte_map: &mut HashMap<u64, TempCellInfo, ahash::RandomState>,
) -> usize {
let mut tbuf = vec![0u8; 65536];
let mut total_bytes = 0usize;
let header_size = 2 * std::mem::size_of::<u32>() as u64;
let size_of_u32 = std::mem::size_of::<u32>();
let size_of_bc = bct.bytes_for_type();
let size_of_umi = umit.bytes_for_type();
let calc_record_bytes = |num_aln: usize| -> usize {
size_of_u32 + size_of_bc + size_of_umi + (size_of_u32 * num_aln)
};
// read each record
for _ in 0..(nrec as usize) {
// read the header of the record
// we don't bother reading the whole thing here
// because we will just copy later as need be
let tup = AlevinFryReadRecord::from_bytes_record_header(reader, bct, umit);
// get the entry for this chunk, or create a new one
let v = cb_byte_map.entry(tup.0).or_insert(TempCellInfo {
offset: header_size,
nbytes: header_size as u32,
nrec: 0_u32,
});
// read the alignment records from the input file
let na = tup.2 as usize;
let req_size = size_of_u32 * na;
if tbuf.len() < req_size {
tbuf.resize(req_size, 0);
}
reader.read_exact(&mut tbuf[0..(size_of_u32 * na)]).unwrap();
// compute the total number of bytes this record requires
let nbytes = calc_record_bytes(na);
v.offset += nbytes as u64;
v.nbytes += nbytes as u32;
v.nrec += 1;
total_bytes += nbytes;
}
// each cell will have a header (8 bytes each)
total_bytes += cb_byte_map.len() * header_size as usize;
let mut output_buffer = Cursor::new(vec![0u8; total_bytes]);
// loop over all distinct cell barcodes, write their
// corresponding chunk header, and compute what the
// offset in `output_buffer` is where the corresponding
// records should start.
let mut next_offset = 0u64;
for (_, v) in cb_byte_map.iter_mut() {
// jump to the position where this chunk should start
// and write the header
output_buffer.set_position(next_offset);
let cell_bytes = v.nbytes;
let cell_rec = v.nrec;
output_buffer.write_all(&cell_bytes.to_le_bytes()).unwrap();
output_buffer.write_all(&cell_rec.to_le_bytes()).unwrap();
// where we will start writing records for this cell
v.offset = output_buffer.position();
// the number of bytes allocated to this chunk
let nbytes = v.nbytes as u64;
// the next record will start after this one
next_offset += nbytes;
}
// now each key points to where we should write the next record for the CB
// reset the input pointer
reader
.get_mut()
.seek(SeekFrom::Start(0))
.expect("could not get read pointer.");
// for each record, read it
for _ in 0..(nrec as usize) {
// read the header of the record
// we don't bother reading the whole thing here
// because we will just copy later as need be
let tup = AlevinFryReadRecord::from_bytes_record_header(reader, bct, umit);
// get the entry for this chunk, or create a new one
if let Some(v) = cb_byte_map.get_mut(&tup.0) {
output_buffer.set_position(v.offset);
// write the num align
let na = tup.2 as usize;
let nau32 = na as u32;
output_buffer.write_all(&nau32.to_le_bytes()).unwrap();
// write the corrected barcode
bct.write_to(tup.0, &mut output_buffer).unwrap();
umit.write_to(tup.1, &mut output_buffer).unwrap();
// read the alignment records
reader.read_exact(&mut tbuf[0..(size_of_u32 * na)]).unwrap();
// write them
output_buffer
.write_all(&tbuf[..(size_of_u32 * na)])
.unwrap();
v.offset = output_buffer.position();
} else {
panic!("should not have any barcodes we can't find");
}
}
output_buffer.set_position(0);
if compress {
// compress the contents of output_buffer to compressed_output
let mut compressed_output =
snap::write::FrameEncoder::new(Cursor::new(Vec::<u8>::with_capacity(total_bytes)));
compressed_output
.write_all(output_buffer.get_ref())
.expect("could not compress the output chunk.");
output_buffer = compressed_output
.into_inner()
.expect("couldn't unwrap the FrameEncoder.");
output_buffer.set_position(0);
}
owriter
.lock()
.unwrap()
.write_all(output_buffer.get_ref())
.unwrap();
cb_byte_map.len()
}
pub fn collate_temporary_bucket_twopass_atac<T: Read + Seek, U: Write>(
reader: &mut BufReader<T>,
bct: &RadIntId,
nrec: u32,
owriter: &Mutex<U>,
compress: bool,
cb_byte_map: &mut HashMap<u64, TempCellInfo, ahash::RandomState>,
) -> usize {
let mut tbuf = vec![0u8; 65536];
let mut total_bytes = 0usize;
let header_size = 2 * std::mem::size_of::<u32>() as u64;
let size_of_bc = bct.bytes_for_type();
let size_of_rec = std::mem::size_of::<u32>()
+ std::mem::size_of::<u8>()
+ std::mem::size_of::<u32>()
+ std::mem::size_of::<u16>();
let size_of_u32 = std::mem::size_of::<u32>();
let calc_record_bytes =
|num_aln: usize| -> usize { size_of_u32 + size_of_bc + (size_of_rec * num_aln) };
// read each record
for _ in 0..(nrec as usize) {
// read the header of the record
// we don't bother reading the whole thing here
// because we will just copy later as need be
let tup = AtacSeqReadRecord::from_bytes_record_header(reader, bct);
// get the entry for this chunk, or create a new one
let v = cb_byte_map.entry(tup.0).or_insert(TempCellInfo {
offset: header_size,
nbytes: header_size as u32,
nrec: 0_u32,
});
// read the alignment records from the input file
let na = tup.1 as usize;
let req_size = size_of_rec * na;
if tbuf.len() < req_size {
tbuf.resize(req_size, 0);
}
reader.read_exact(&mut tbuf[0..(size_of_rec * na)]).unwrap();
// compute the total number of bytes this record requires
let nbytes = calc_record_bytes(na);
v.offset += nbytes as u64;
v.nbytes += nbytes as u32;
v.nrec += 1;
total_bytes += nbytes;
}
// each cell will have a header (8 bytes each)
total_bytes += cb_byte_map.len() * header_size as usize;
let mut output_buffer = Cursor::new(vec![0u8; total_bytes]);
// loop over all distinct cell barcodes, write their
// corresponding chunk header, and compute what the
// offset in `output_buffer` is where the corresponding
// records should start.
let mut next_offset = 0u64;
for (_, v) in cb_byte_map.iter_mut() {
// jump to the position where this chunk should start
// and write the header
output_buffer.set_position(next_offset);
let cell_bytes = v.nbytes;
let cell_rec = v.nrec;
output_buffer.write_all(&cell_bytes.to_le_bytes()).unwrap();
output_buffer.write_all(&cell_rec.to_le_bytes()).unwrap();
// where we will start writing records for this cell
v.offset = output_buffer.position();
// the number of bytes allocated to this chunk
let nbytes = v.nbytes as u64;
// the next record will start after this one
next_offset += nbytes;
}
// now each key points to where we should write the next record for the CB
// reset the input pointer
reader
.get_mut()
.seek(SeekFrom::Start(0))
.expect("could not get read pointer.");
// for each record, read it
for _ in 0..(nrec as usize) {
// read the header of the record
// we don't bother reading the whole thing here
// because we will just copy later as need be
let tup = AtacSeqReadRecord::from_bytes_record_header(reader, bct);
// get the entry for this chunk, or create a new one
if let Some(v) = cb_byte_map.get_mut(&tup.0) {
output_buffer.set_position(v.offset);
// write the num align
let na = tup.1 as usize;
let nau32 = na as u32;
output_buffer.write_all(&nau32.to_le_bytes()).unwrap();
// write the corrected barcode
bct.write_to(tup.0, &mut output_buffer).unwrap();
// read the alignment records
reader.read_exact(&mut tbuf[0..(size_of_rec * na)]).unwrap();
// write them
output_buffer
.write_all(&tbuf[..(size_of_rec * na)])
.unwrap();
v.offset = output_buffer.position();
} else {
panic!("should not have any barcodes we can't find");
}
}
output_buffer.set_position(0);
if compress {
// compress the contents of output_buffer to compressed_output
let mut compressed_output =
snap::write::FrameEncoder::new(Cursor::new(Vec::<u8>::with_capacity(total_bytes)));
compressed_output
.write_all(output_buffer.get_ref())
.expect("could not compress the output chunk.");
output_buffer = compressed_output
.into_inner()
.expect("couldn't unwrap the FrameEncoder.");
output_buffer.set_position(0);
}
owriter
.lock()
.unwrap()
.write_all(output_buffer.get_ref())
.unwrap();
cb_byte_map.len()
}
/*
pub fn collate_temporary_bucket<T: Read>(
reader: &mut T,
bct: &RadIntId,
umit: &RadIntId,
_nchunks: u32,
nrec: u32,
output_cache: &mut HashMap<u64, CorrectedCbChunk, ahash::RandomState>,
) {
let mut tbuf = [0u8; 65536];
// estimated average number of records per barcode
// this is just for trying to pre-allocate buffers
// right; should not affect correctness
let est_num_rec = 1; //(nrec / nchunks) + 1;
// for each record, read it
for _ in 0..(nrec as usize) {
// read the header of the record
// we don't bother reading the whole thing here
// because we will just copy later as need be
let tup = AlevinFryReadRecord::from_bytes_record_header(reader, bct, umit);
// get the entry for this chunk, or create a new one
let v = output_cache
.entry(tup.0)
.or_insert_with(|| CorrectedCbChunk::from_label_and_counter(tup.0, est_num_rec));
// keep track of the number of records we're writing
v.nrec += 1;
// write the num align
let na = tup.2;
v.data.write_all(&na.to_le_bytes()).unwrap();
// write the corrected barcode
bct.write_to(tup.0, &mut v.data).unwrap();
umit.write_to(tup.1, &mut v.data).unwrap();
// read the alignment records
reader.read_exact(&mut tbuf[0..(4 * na as usize)]).unwrap();
// write them
v.data.write_all(&tbuf[..(4 * na as usize)]).unwrap();
}
}
pub fn process_corrected_cb_chunk<T: Read>(
reader: &mut T,
bct: &RadIntId,
umit: &RadIntId,
correct_map: &HashMap<u64, u64>,
expected_ori: &Strand,
output_cache: &DashMap<u64, CorrectedCbChunk>,
owriter: &Mutex<BufWriter<File>>,
) {
let mut buf = [0u8; 8];
let mut tbuf = [0u8; 65536];
// get the number of bytes and records for
// the next chunk
reader.read_exact(&mut buf).unwrap();
let _nbytes = buf.pread::<u32>(0).unwrap();
let nrec = buf.pread::<u32>(4).unwrap();
// for each record, read it
for _ in 0..(nrec as usize) {
let tup = AlevinFryReadRecord::from_bytes_record_header(reader, bct, umit);
// if this record had a correct or correctable barcode
if let Some(corrected_id) = correct_map.get(&tup.0) {
let rr = AlevinFryReadRecord::from_bytes_with_header_keep_ori(
reader,
tup.0,
tup.1,
tup.2,
expected_ori,
);
if let Some(mut v) = output_cache.get_mut(corrected_id) {
// update the corresponding corrected chunk entry
v.remaining_records -= 1;
let last_record = v.remaining_records == 0;
// if there are no alignments in the record
// (potentially b/c of orientation filtering)
// then don't push info on to the vector.
if rr.is_empty() {
if last_record {
dump_chunk(&mut v, owriter);
}
continue;
}
v.nrec += 1;
let na = rr.refs.len() as u32;
v.data.write_all(&na.to_le_bytes()).unwrap();
bct.write_to(*corrected_id, &mut v.data).unwrap();
umit.write_to(rr.umi, &mut v.data).unwrap();
v.data.write_all(as_u8_slice(&rr.refs[..])).unwrap();
if last_record {
dump_chunk(&mut v, owriter);
}
}
} else {
reader
.read_exact(&mut tbuf[0..(4 * (tup.2 as usize))])
.unwrap();
}
}
}
*/
/// Represents a temporary bucket of barcodes whose records will
/// be written together and then collated later in memory.
pub struct TempBucket {
pub bucket_id: u32,
pub bucket_writer: Arc<Mutex<BufWriter<File>>>,
pub num_chunks: u32,
pub num_records: u32,
pub num_records_written: AtomicU32,
pub num_bytes_written: AtomicU64,
}
impl TempBucket {
pub fn from_id_and_parent(bucket_id: u32, parent: &std::path::Path) -> Self {
TempBucket {
bucket_id,
bucket_writer: Arc::new(Mutex::new(BufWriter::with_capacity(
4096_usize,
File::create(parent.join(format!("bucket_{}.tmp", bucket_id))).unwrap(),
))),
num_chunks: 0u32,
num_records: 0u32,
num_records_written: AtomicU32::new(0u32),
num_bytes_written: AtomicU64::new(0u64),
}
}
}
/// Read an input chunk from `reader` and write the
/// resulting records to the corresponding in-memory
/// buffers `local_buffers`. As soon as any buffer
/// reaches `flush_limit`, flush the buffer by writing
/// it to the `output_cache`.
#[allow(clippy::too_many_arguments)]
pub fn dump_corrected_cb_chunk_to_temp_file<T: Read>(
reader: &mut BufReader<T>,
bct: &RadIntId,
umit: &RadIntId,
correct_map: &HashMap<u64, u64>,
expected_ori: &Strand,
output_cache: &HashMap<u64, Arc<TempBucket>>,
local_buffers: &mut [Cursor<&mut [u8]>],
flush_limit: usize,
) {
let mut buf = [0u8; 8];
let mut tbuf = vec![0u8; 4096];
//let mut tcursor = Cursor::new(tbuf);
//tcursor.set_position(0);
// get the number of bytes and records for
// the next chunk
reader.read_exact(&mut buf).unwrap();
let _nbytes = buf.pread::<u32>(0).unwrap();
let nrec = buf.pread::<u32>(4).unwrap();
let bc_bytes = bct.bytes_for_type();
let umi_bytes = umit.bytes_for_type();
let na_bytes = std::mem::size_of::<u32>();
let target_id_bytes = std::mem::size_of::<u32>();
// for each record, read it
for _ in 0..(nrec as usize) {
let tup = AlevinFryReadRecord::from_bytes_record_header(reader, bct, umit);
// if this record had a correct or correctable barcode
if let Some(corrected_id) = correct_map.get(&tup.0) {
let rr = AlevinFryReadRecord::from_bytes_with_header_keep_ori(
reader,
tup.0,
tup.1,
tup.2,
expected_ori,
);
if rr.is_empty() {
continue;
}
if let Some(v) = output_cache.get(corrected_id) {
// if this is a valid barcode, then
// write the corresponding entry to the
// thread-local buffer for this bucket
// the total number of bytes this record will take
let nb = (rr.refs.len() * target_id_bytes + na_bytes + bc_bytes + umi_bytes) as u64;
// the buffer index for this corrected barcode
let buffidx = v.bucket_id as usize;
// the current cursor for this buffer
let bcursor = &mut local_buffers[buffidx];
// the current position of the cursor
let len = bcursor.position() as usize;
// if writing the next record (nb bytes) will put us over
// the flush size for the thread-local buffer for this bucket
// then first flush the buffer to file.
if len + nb as usize >= flush_limit {
let mut filebuf = v.bucket_writer.lock().unwrap();
filebuf.write_all(&bcursor.get_ref()[0..len]).unwrap();
// and reset the local buffer cursor
bcursor.set_position(0);
}
// now, write the record to the buffer
let na = rr.refs.len() as u32;
bcursor.write_all(&na.to_le_bytes()).unwrap();
bct.write_to(*corrected_id, bcursor).unwrap();
umit.write_to(rr.umi, bcursor).unwrap();
bcursor.write_all(as_u8_slice(&rr.refs[..])).unwrap();
// update number of written records
v.num_records_written.fetch_add(1, Ordering::SeqCst);
// update number of written bytes
v.num_bytes_written.fetch_add(nb, Ordering::SeqCst);
}
} else {
// in this branch, we don't have access to a correct barcode for
// what we observed, so we need to discard the remaining part of
// the record.
let req_len = target_id_bytes * (tup.2 as usize);
let do_resize = req_len > tbuf.len();
if do_resize {
tbuf.resize(req_len, 0);
}
reader
.read_exact(&mut tbuf[0..(target_id_bytes * (tup.2 as usize))])
.unwrap();
if do_resize {
tbuf.resize(4096, 0);
tbuf.shrink_to_fit();
}
}
}
}
/// Read an input chunk from `reader` and write the
/// resulting records to the corresponding in-memory
/// buffers `local_buffers`. As soon as any buffer
/// reaches `flush_limit`, flush the buffer by writing
/// it to the `output_cache`.
#[allow(clippy::too_many_arguments)]
pub fn dump_corrected_cb_chunk_to_temp_file_atac<T: Read>(
reader: &mut BufReader<T>,
bct: &RadIntId,
correct_map: &HashMap<u64, u64>,
output_cache: &HashMap<u64, Arc<TempBucket>>,
local_buffers: &mut [Cursor<&mut [u8]>],
flush_limit: usize,
) {
let mut buf = [0u8; 8];
let mut tbuf = vec![0u8; 4096];
//let mut tcursor = Cursor::new(tbuf);
//tcursor.set_position(0);
// get the number of bytes and records for
// the next chunk
reader.read_exact(&mut buf).unwrap();
let _nbytes = buf.pread::<u32>(0).unwrap();
let nrec = buf.pread::<u32>(4).unwrap();
let bc_bytes = bct.bytes_for_type();
let na_bytes = std::mem::size_of::<u32>();
let target_id_bytes = std::mem::size_of::<u32>() + // ref id
std::mem::size_of::<u8>() + // type
std::mem::size_of::<u32>() + // position
std::mem::size_of::<u16>(); // frag_len
// for each record, read it
for _ in 0..(nrec as usize) {
let tup = AtacSeqReadRecord::from_bytes_record_header(reader, bct);
// if this record had a correct or correctable barcode
if let Some(corrected_id) = correct_map.get(&tup.0) {
// could be replaced with orientation
let rr = AtacSeqReadRecord::from_bytes_with_header(reader, tup.0, tup.1);
if rr.is_empty() {
continue;
}
if let Some(v) = output_cache.get(corrected_id) {
// if this is a valid barcode, then
// write the corresponding entry to the
// thread-local buffer for this bucket
// the total number of bytes this record will take
let nb = (rr.refs.len() * target_id_bytes + na_bytes + bc_bytes) as u64;
// the buffer index for this corrected barcode
let buffidx = v.bucket_id as usize;
// the current cursor for this buffer
let bcursor = &mut local_buffers[buffidx];
// the current position of the cursor
let len = bcursor.position() as usize;
// if writing the next record (nb bytes) will put us over
// the flush size for the thread-local buffer for this bucket
// then first flush the buffer to file.
if len + nb as usize >= flush_limit {
let mut filebuf = v.bucket_writer.lock().unwrap();
filebuf.write_all(&bcursor.get_ref()[0..len]).unwrap();
// and reset the local buffer cursor
bcursor.set_position(0);
}
// now, write the record to the buffer
let na = rr.refs.len() as u32;
bcursor.write_all(&na.to_le_bytes()).unwrap();
bct.write_to(*corrected_id, bcursor).unwrap();
bcursor.write_all(as_u8_slice(&rr.refs[..])).unwrap();
bcursor.write_all(as_u8_slice_u8(&rr.map_type[..])).unwrap();
bcursor.write_all(as_u8_slice(&rr.start_pos[..])).unwrap();
bcursor
.write_all(as_u8_slice_u16(&rr.frag_lengths[..]))
.unwrap();
// update number of written records
v.num_records_written.fetch_add(1, Ordering::SeqCst);
// update number of written bytes
v.num_bytes_written.fetch_add(nb, Ordering::SeqCst);
}
} else {
// in this branch, we don't have access to a correct barcode for
// what we observed, so we need to discard the remaining part of
// the record.
let req_len = target_id_bytes * (tup.1 as usize);
let do_resize = req_len > tbuf.len();
if do_resize {
tbuf.resize(req_len, 0);
}
reader
.read_exact(&mut tbuf[0..(target_id_bytes * (tup.1 as usize))])
.unwrap();
if do_resize {
tbuf.resize(4096, 0);
tbuf.shrink_to_fit();
}
}
}
}
pub fn as_u8_slice(v: &[u32]) -> &[u8] {
unsafe { std::slice::from_raw_parts(v.as_ptr() as *const u8, std::mem::size_of_val(v)) }
}
pub fn as_u8_slice_u16(v: &[u16]) -> &[u8] {
unsafe { std::slice::from_raw_parts(v.as_ptr() as *const u8, std::mem::size_of_val(v)) }
}
pub fn as_u8_slice_u8(v: &[u8]) -> &[u8] {
unsafe { std::slice::from_raw_parts(v.as_ptr(), std::mem::size_of_val(v)) }
}
#[cfg(test)]
mod tests {
use crate::BarcodeLookupMap;
use needletail;
#[test]
fn test_barcode_lookup_map() {
let barcode_sv_even = vec![
b"AACC", b"AAGG", b"CAGT", b"CATT", b"GACC", b"GATA", b"TCAG", b"TCGT",
];
let barcode_sv_odd = vec![
b"AACCA", b"AAGGC", b"CAGTA", b"CATTG", b"GACCG", b"GATAC", b"TCAGA", b"TCGTG",
];
let mut barcode_even = Vec::with_capacity(barcode_sv_even.len());
for b in barcode_sv_even.clone() {
if let Some((_, km, _)) =
needletail::bitkmer::BitNuclKmer::new(&b[..], b.len() as u8, false).next()
{
barcode_even.push(km.0);
}
}
let mut barcode_odd = Vec::with_capacity(barcode_sv_odd.len());
for b in barcode_sv_odd.clone() {
if let Some((_, km, _)) =
needletail::bitkmer::BitNuclKmer::new(&b[..], b.len() as u8, false).next()
{
barcode_odd.push(km.0);
}
}
let me = BarcodeLookupMap::new(barcode_even, 4);
let mo = BarcodeLookupMap::new(barcode_odd, 5);
let x = b"CAGA";
if let Some((_, et, _)) =
needletail::bitkmer::BitNuclKmer::new(&x[..], x.len() as u8, false).next()
{
assert_eq!((Some(2), 1), me.find_neighbors(et.0, false));
}
let x = b"CAATG";
if let Some((_, et, _)) =
needletail::bitkmer::BitNuclKmer::new(&x[..], x.len() as u8, false).next()
{
assert_eq!((Some(3), 1), mo.find_neighbors(et.0, false));
}
}
}