Skip to content

Instantly share code, notes, and snippets.

@kloetzl

kloetzl/reaper.c

Created July 14, 2017 07:23
Show Gist options
  • Save kloetzl/1dd24925c031d85fc9c3222c4aed8b00 to your computer and use it in GitHub Desktop.
Save kloetzl/1dd24925c031d85fc9c3222c4aed8b00 to your computer and use it in GitHub Desktop.
Download ZIP
reaper source
Raw
reaper.c
/*
* (C) Copyright 2011, 2012, 2013, 2014, 2015 European Molecular Biology Laboratory.
* Author: Stijn van Dongen <stijn@ebi.ac.uk>.
* Contact: <kraken@ebi.ac.uk>
*
* This file is part of Reaper. Reaper is free software: you can redistribute
* it and/or modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; either version 3 of the License,
* or (at your option) any later version. This program is distributed in the
* hope that it will be useful, but WITHOUT ANY WARRANTY; without even the
* implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
* See the GNU General Public License for more details. You should have
* received a copy of the GNU General Public License along with this program.
* If not, see http://www.gnu.org/licenses/.
*/
/*
This is reaper, a flexible and fast program for pre-processing short-read sequencing data.
- Match read with 3p-adaptor
- uses Smith-Waterman local alignment over full read and adaptor
1. The best local alignment is considered and tested against user-set thresholds
2. Optionally weaker matches between read-suffix and adaptor-prefix can be considered as well
- read/adaptor match is user controlled (alignment identity threshold)
- handling of poly-A (optional)
- handling of N in sequence (optional)
- handling of B quality scores (optional)
- Check for ligation contamination
- Smith-Waterman local alignment
- Handles barcodes
- Output split into per-barcode clean files and annotated discard files
- Recognises different layouts of barcodes and sequence inserts
- Sane defaults and adaptable.
- Outputs per-barcode summary statistics for comprehensive QC plots
- Output fields can be chosen and formatted
- Per-read characteristics available (e.g. alignment, trinucleotide score)
- One read-at-time, written in C
- Small memory footprint (order of megabytes)
- 1-4 million reads per minute (depending on read length, contamination check)
- Twinned with 'tally' for fast uniquifying of reads
NOTES
* Implementation caveat I.
param->buck[buck_n] exists, contrary to normal C convention.
qc_tally_in and qc_tally_out write in this bucket - it is used for
grand-total reporting. See /specialbucket/.
This has the advantage that normal bucket processing code looks normal, and
will not screw up the special bucket - only memory management looks peculiar.
* Implementation caveat II.
The code mixes 0-based ofsets and 1-based ofsets. The first derive
from normal C-strings, the second from alignment matrices -- this
is because they are zero-bordered; the first base in e.g. the adaptor
maps to the second column of the alignment matrix, indexed 1.
See /stringoffsets/.
* Implementation Note.
do_adaptor_3p and do_barcode_3p are quite different. The former allows the
absence of a match, the latter requires exactly one match. The former allows
more laxness when having an end-to-start match, and additionally allows a
3p-adaptor prefix match if it is folllowed by low complexity code.
*/
#define REAPER_FAIL 29
#define REAPER_OK 0
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdarg.h>
#include <signal.h>
#include <limits.h>
#include <time.h>
#include <zlib.h>
#include <ctype.h>
#include "reaper.h"
#include "version.h"
#include "sw.h"
#include "slib.h"
#include "trint.h"
#include "table.h"
#include "dna.h"
#ifndef MAXFIELDSIZE
# define MAXFIELDSIZE 1023
#endif
int g_debug = 0;
int g_guard = 0;
int g_restrict = 18;
unsigned long g_fieldtosmall = 0;
/* NOTE these are coupled to the below array */
enum program_mode
{ CASE_NOTSPECIFIED = 0
/* Caveat: we use the test g_mode > CASE_NOBC in the program */
, CASE_NOBC
, CASE_3P_BC
, CASE_5P_BC
} ;
/* NOTE these are coupled to the above enum */
const char* G_MODE[6]
= { "NA"
, "no-bc"
, "3p-bc"
, "5p-bc"
};
enum program_mode g_mode = CASE_NOTSPECIFIED;
const char* x_3pgb[2] = { "3p-global -", "3p-global +" };
const char* x_tabu[2] = { "tabu -", "tabu +" };
const char* x_3ppx[2] = { "3p-prefix -", "3p-prefix +" };
const char* x_3pbc[2] = { "3p-barcode -", "3p-barcode +" };
const char* x_5pbc[2] = { "5p-barcode -", "5p-barcode +" };
const char* x_5psi[2] = { "3p-sinsert -", "3p-sinsert +" };
int dna[5] = { 'A', 'C', 'G', 'T', 'N' };
#define MATRIX_SIZE 8192
#define MAXQUALITY 256
#define dim unsigned
struct qc_composition
{ dim A[MAXFIELDSIZE]
; dim C[MAXFIELDSIZE]
; dim G[MAXFIELDSIZE]
; dim T[MAXFIELDSIZE]
; dim N[MAXFIELDSIZE]
;
} ;
struct atabu
{ char* tabuseq
; unsigned tabulen
; struct match_requirement mr_tabu
; struct atabu* next
;
} ;
/* Initially only used for barcode funneling, but can also be used for
* other purposes. This bucket design with multiple purposes does not
* yet have a transparent design: the last bucket in struct param is
* the only non-barcode bucket, and is special.
* - barcodes_io_open only opens streams for the barcode buckets.
*/
struct bucket
{ const char* barcode3p
; const char* sinsert3p
; const char* adaptor3p
; const char* barcode5p
; const char* sinsert5p
; const char* cat3p /* new */
; int sinsert3p_n /* new */
; int sinsert5p_n /* new */
; const char* filetag /* use instead of barcode in file name */
; struct atabu tabu2 /* fixme, linked list with first link not freeable, designtabu
* The '2' in tabu2 has no meaning -- artefact from rewrite,
* kept around because it makes searching easier.
* Too lazy to think of something better.
*/
; int barcode_n
; int barcode_start
; int barcode_end
; int cache_lint_offset
; int cache_score
; int cache_accept
; char* fname
; struct qc_composition qc_cpsin /* input read composition */
; struct qc_composition qc_cpsout /* clean read composition */
; unsigned global_in_max_size
; unsigned global_out_max_size
; dim qc_lenout[MAXFIELDSIZE+1] /* clean read length */
; dim * qc_bcq /* base call quality, one block is one read offet
* read-offset 0: q0, q1, q2 ..... q255
* read-offset 1: q0, q1, q2 ..... q255
* .....
* read-offset N-1: q0, q1, q2 ..... q255
*/
; ZFILE fp
;
} ;
struct match_account
{ unsigned match_type /* n(one) g(lobal) p(refix), t(ail) */
; double suffix_complexity /* for prefix match */
; unsigned n_perfect /* head to tail length */
; int nnn_offset
; int bbb_offset
; int aaa_offset
; int qqq_offset
; int dust_offset
;
} ;
struct param
{ unsigned cleanlength
#define VB_ALIGN 1 << 0
#define VB_LINT 1 << 1
#define VB_CLEAN 1 << 2
; unsigned verbose
#define MODE_QC_EARLY 1 << 0
#define MODE_QC_LATE 1 << 1
#define MODE_QC_FULL 1 << 3
#define MODE_FULLLENGTH 1 << 4
#define MODE_KEEPALL 1 << 5
#define MODE_DUSTPOLYA 1 << 6
#define MODE_STREAMCLEAN 1 << 7
; unsigned modes
; unsigned sample_lint
; unsigned sample_clean
/* we kludge/fudge this a bit. buck[buck_n] will also be utilised
* contrary to common conventions. It's an 'overflow' bucket
* for all reads that could not be classified in the barcode cases,
* or just all reads in the case where no barcode is present.
*/
; struct bucket *buck
; int buck_n /* will be -1 initalised, so that loops 0 <= buck_n work */
; unsigned p_3p_tth /* prefix match, minimal tail to head perfect match */
; const char* fnmeta
; ZFILE fpclean
; ZFILE fplint
; int zippit
; int polya /* strip poly-As of at least this size */
; unsigned taildust /* strip tail if dust-like criterion exceeds this */
; unsigned taildustlate /* after adapter matching, strip tail if dust-like criterion exceeds this */
; unsigned taildust_bases
; unsigned taildustlate_bases
; int trintco /* if != 0, remove reads with tri-nt score >= trintco */
; int length_co
; int q_winlen
; int q_winofs
; int q_readofs /* skip first readofs bases */
; int q_cutoff
; const char* adaptor_3p
; const char* sinsert_5p
; const char* adaptor_test /* experimental */
; struct atabu the_tabu2
; const char* basename
; struct sw_param swp_3p
; struct sw_param swp_5p
; struct sw_param swp_tabu
; struct sw_param swp_raptor /* reaper - adaptor testing */
; struct match_requirement mr_3p_global /* used without clause */
; struct match_requirement mr_3p_prefix /* only used if mr_minlen > 0 */
; struct match_requirement mr_3p_barcode /* used without clause */
; struct match_requirement mr_5p_require /* used without clause */
; struct match_requirement mr_tabu /* used without clause */
; struct match_requirement mr_5p_barcode
; double suffix_complexity_max /* in conjunction with prefix match */
; double loco_check
; unsigned nnn_count
; unsigned nnn_outof
; int anchor_offset
; int length_limit
; const char* anchor_string
; unsigned kmer /* to analyse distance to adapter */
; const char* format_clean
; const char* format_lint
; struct table metadata
;
} ;
void param_init
( struct param * pam
) /* 1. minlen
| 2. maxedit
| | 3. maxgap
| | | 4. offset
| | | | */
{ struct match_requirement mr_3p_global = { 14, 2, 1, 0 }
; struct match_requirement mr_3p_prefix = { 8, 2, 0, 2 }
; struct match_requirement mr_3p_barcode= { 0, 6, 1, 0 }
; struct match_requirement mr_5p_require= { 0, 2, 1, 10}
; struct match_requirement mr_tabu = { 16, 2, 1, 0 }
; struct match_requirement mr_5p_barcode= { 0, 0, 0, 2 } /* bit field 1 = zip 2 = normal alignment */
; struct atabu theboo = { "", 0, { 0 }, NULL }
; pam->cleanlength = 0
; pam->verbose = 0
; pam->modes = MODE_QC_FULL
; pam->sample_clean = 1
; pam->sample_lint = 1
; pam->polya = 0
; pam->taildust = 0
; pam->taildustlate = 0
; pam->taildust_bases = 0
; pam->taildustlate_bases = 0
; pam->p_3p_tth = 0 /* strip perfect (read) tail to head (3p adaptor) match of any length >= 1 */
; pam->mr_3p_global = mr_3p_global
; pam->mr_3p_prefix = mr_3p_prefix
; pam->mr_3p_barcode = mr_3p_barcode
; pam->mr_5p_require = mr_5p_require
; pam->mr_tabu = mr_tabu
; pam->mr_5p_barcode = mr_5p_barcode
; pam->buck = NULL
; pam->buck_n = -1
; pam->fnmeta = NULL
; pam->adaptor_3p = NULL
; pam->adaptor_test = NULL
; pam->sinsert_5p = NULL
; pam->basename = "out"
; pam->trintco = 0
; pam->length_co = 0
; pam->q_winlen = 0
; pam->q_winofs = 0
; pam->q_readofs = 10
; pam->q_cutoff = 0
; pam->zippit =
#if WE_USE_ZLIB
1
#else
0
#endif
; pam->swp_3p.cost_gapleft = 3
; pam->swp_3p.cost_gapright = 3
; pam->swp_3p.cost_subst = 1
; pam->swp_3p.gain_match = 4
; pam->swp_3p.left_skip = 0
; pam->swp_3p.right_limit = 0
; pam->swp_3p.flags = 0
; pam->swp_5p = pam->swp_3p
; pam->swp_tabu = pam->swp_3p
; pam->swp_raptor = pam->swp_3p
; pam->swp_raptor.cost_gapleft = 1
; pam->swp_raptor.cost_gapright = 1
; pam->swp_raptor.cost_subst = 1
; pam->swp_raptor.gain_match = 2
; pam->the_tabu2 = theboo
; pam->nnn_count = 0
; pam->nnn_outof = 0
; pam->anchor_offset = -1
; pam->anchor_string = NULL
; pam->kmer = 10
; pam->suffix_complexity_max = 0.25
; pam->loco_check = 0.0
; pam->format_clean = "@%I%n%C%n+%n%Q%n"
; pam->format_lint = "@%I%tmsg=%M%n%R%n+%n%Y%n"
; pam->length_limit = 0
; pam->fpclean = NULL
; pam->fplint = NULL
; memset(&pam->metadata, 0, sizeof pam->metadata)
; }
unsigned get_mdta /* minimum distance to adapter, per k-mer varied along read */
( char* buf_distance
, char* buf_offset
, const char* seq
, unsigned seq_n
, struct param* pam
)
{ const char* theaptr = pam->adaptor_test
; unsigned k = pam->kmer
; struct sw_param* swp = &pam->swp_raptor
; static unsigned char* mdta = NULL
; unsigned mask = 0
; unsigned kmer = 0
; char seq2[MAXFIELDSIZE+1]
; int i = 0
; if (!mdta)
mdta = calloc(1 << 2*k, 1)
; buf_distance[0] = '\0'
; buf_offset[0] = '\0'
; strncpy(seq2, seq, MAXFIELDSIZE)
; if (!theaptr)
theaptr = pam->adaptor_3p
; if (!theaptr) /* give up */
seq_n = 0
; for (i=0; i < seq_n; i++)
{ int dist = 7, offset = 31
; unsigned b = themap[(unsigned char) seq2[i]]
; mask = (mask << 1) & ((1 << k) -1)
; kmer = (kmer << 2) & ((1 << 2*k) -1)
; if (b > 3)
mask |= 1
; else
kmer |= b
;if(0)fprintf(stderr, "[%s]\n", seq2+i)
; if (i+1 < k)
{ buf_distance[i] = '.'
; buf_distance[seq_n+i+1] = '.'
; continue
; }
/* store distance in 3 bits (0..7), offset in 5 bits (1..31) */
if (mask || !mdta[kmer])
{ SWNUM data[MATRIX_SIZE] = { 0 }
; struct sw_alninfo ai = { 0 }
; char bu = seq2[i+1]
; seq2[i+1] = '\0' /* sw_fill2 wants 0-terminated. blah. */
; if
( !sw_fill2
( &ai
, data
, MATRIX_SIZE
, seq2 + i+1-k
, theaptr
, swp->cost_gapleft + swp->cost_gapright
, swp
) )
{ sw_trace3(&ai, swp, ai.max_ij, 0) /* do_barcode_3p */
; dist = k - ai.n_match + ai.n_insr + ai.n_insl
; offset = ai.rgt_start
; }
if (dist > 7) dist = 7
; if (offset > 31) offset = 31
; if (!mask)
mdta[kmer] = dist | (offset << 3)
; seq2[i+1] = bu
; }
else
dist = mdta[kmer] & 7
, offset = mdta[kmer] >> 3
; buf_distance[i] = '0' + (dist > 9 ? 9 : dist)
; buf_distance[seq_n+i+1] = offset["0123456789ABCDEFGHIJKLMNOPQRSTUV"]
; }
; if (seq_n)
{ buf_distance[i] = '\n'
; buf_distance[2 * seq_n+1] = '\0'
; buf_distance[0] = 'D'
; buf_distance[seq_n+1] = 'O'
; }
else
{ memcpy(buf_distance, "D\nO\0", 4)
; return 3
; }
return 2 * seq_n + 1
; }
struct read_stats_global
{ unsigned N
; unsigned N_err /* globally set */
; unsigned N_truncated /* */
; unsigned N_alien /* not ACGTUNX */
; unsigned N_cflr /* conflict resolved, per case */
; unsigned N_clean /* per case */
; unsigned N_bb /* globally set */
; unsigned N_qq /* globally set */
; unsigned N_anchor /* globally set */
; unsigned N_loco /* globally set */
; unsigned N_nomsg /* globally set */
; unsigned N_cfl /* per case */
; unsigned N_nn /* per case */
; unsigned N_aa /* per case */
; unsigned N_tri /* per case */
; unsigned N_nomatch /* per case */
; unsigned N_lenlo /* per case */
; unsigned N_tabu /* per case */
; unsigned N_5p_nosinsert /* per case */ /* This group should add up to */
; unsigned N_lint /* per case */ /* to this */
; unsigned long B /* */
; unsigned long N_cells /* total size of alignment matrices */
; unsigned B_BB /* bases lost due to BBB, globally set */
; unsigned B_tri /* bases lost due to tri-tail trimming, globally set */
; unsigned B_NN /* bases lost due to NNN, globally set */
; unsigned B_AA /* bases lost due to AAA, (after adaptor stripping) */
; unsigned B_QQ /* bases lost due to QQQ, (after adaptor stripping) */
; unsigned B_lco /* bases lost due to user specified length cutoff */
/* hierverder; implement below */
; unsigned B_3pA /* bases lost due to 3pA stripping */
; unsigned B_5pMM /* mismatches in required 5p adaptor */
;
} ;
int cpytofield
( char* dest
, char* src
, int n
)
{ if (n > MAXFIELDSIZE)
n = MAXFIELDSIZE
, g_fieldtosmall++
; memcpy(dest, src, n)
; dest[n] = '\0'
; return n
; }
struct record
{ char id [MAXFIELDSIZE+1]
; char seq [MAXFIELDSIZE+1]
; char q [MAXFIELDSIZE+1]
; char discard [MAXFIELDSIZE+1]
; char annot [MAXFIELDSIZE+1]
; unsigned id_n
; unsigned seq_n
; unsigned q_n
; unsigned discard_n
; unsigned annot_n
; int clean_n
; unsigned count
; const char* out_message
; struct read_stats_global rs /* NOTE maintains state between different read_record2 calls */
;
} ;
void validate_sequence
( struct record* rec
)
{ int i
; for (i=0;i<rec->seq_n;i++)
{ rec->seq[i] = toupper((unsigned char) rec->seq[i])
; if (BASEMAP((unsigned char) rec->seq[i]) > 4)
rec->rs.N_alien++
, rec->seq[i] = 'N'
; }
}
#define RR_ERROR 3
#define RR_NOMEM 2
#define RR_DONE 1
#define RR_OK 0
/* At the moment this keeps reading in case of RL_NOMEM,
* but readline does not flush overly long lines yet,
* so there is not yet a mechanism to just ignore
* a record with lines that are too long.
* todo fixme
*/
int read_record2
( ZFILE ip
, const char* format
, struct record* rec
, int length_limit
)
{ unsigned char* a = (unsigned char*) format
; unsigned rlstat = 0
; int q_received = 0
; unsigned truncated = 0
; int maxfieldsize = MAXFIELDSIZE
; if (length_limit && length_limit < MAXFIELDSIZE)
maxfieldsize = length_limit
; rec->id_n = 0
; rec->seq_n = 0
; rec->annot_n = 0
; rec->discard_n = 0
; rec->clean_n = 0
; rec->q_n = 0
; rec->count = 1
; rec->id[0] = '\0'
; rec->seq[0] = '\0'
; rec->discard[0]= '\0'
; rec->q[0] = '\0'
; rec->annot[0] = '\0'
; while (a[0])
{ char* receive = NULL
; unsigned *np_receive = 0
; switch(a[0])
{ case 'I': receive = rec->id; np_receive = &rec->id_n; break
; case 'R': receive = rec->seq; np_receive = &rec->seq_n; break
; case 'D': receive = rec->discard; np_receive = &rec->discard_n; break
; case 'A': receive = rec->annot; np_receive = &rec->annot_n; break
; case 'Q': receive = rec->q; np_receive = &rec->q_n; q_received = 0; break
; }
rlstat |= kraken_readline(ip, receive, maxfieldsize, np_receive, &truncated)
; if (a[0] == 'R')
rec->rs.N_truncated += truncated
; if (rlstat & (RL_DONE | RL_ERROR))
break
; a++
; }
if (rlstat & (RL_DONE | RL_ERROR))
{ if (a != (unsigned char*) format && a[0])
argh("reaper", "last record was incomplete")
; return RR_DONE
; }
if (rlstat & RL_NOMEM) /* fixme not checked yet */
return RR_NOMEM
; rec->rs.N++
; validate_sequence(rec)
; if (q_received && rec->q_n != rec->seq_n)
{ rec->rs.N_err++
; rec->q_n = 0
; rec->q[0] = '\0'
; }
return RR_OK
; }
/*
; case 'I': receive = rec->id; np_receive = &rec->id_n; break
; case 'R': receive = rec->seq; np_receive = &rec->seq_n; break
; case 'D': receive = rec->discard; np_receive = &rec->discard_n; break
; case 'Q': receive = rec->q; np_receive = &rec->q_n; q_received = 0; break
*/
#define izblank(c) ((unsigned char) (c) == ' ' || (unsigned char) (c) == '\t')
int read_recordx
( ZFILE ip
, const char* format
, struct record* rec
)
{ const char* fmtp = format, *fmtz = format + strlen(format)
; unsigned rlstat = 0
#define THE_BUFSIZE 8191
; char buf[THE_BUFSIZE+1]
; unsigned n_received = 0, n_lines = 0
; unsigned truncated = 0
; char* bufp = NULL, *curp = NULL, *bufz = NULL
; int esc = 0
; rec->id_n = 0
; rec->seq_n = 0
; rec->discard_n = 0
; rec->q_n = 0
; rec->clean_n = 0
; rec->count = 1
; rec->id[0] = '\0'
; rec->seq[0] = '\0'
; rec->discard[0]= '\0'
; rec->q[0] = '\0'
; while (fmtp < fmtz)
{ if (!bufp)
{ rlstat |= kraken_readline(ip, buf, THE_BUFSIZE, &n_received, &truncated)
; if (truncated || rlstat & (RL_DONE | RL_ERROR))
break
; bufp = buf
; bufz = buf + n_received
; n_lines++
; esc = 0
; }
curp = bufp
; if (esc)
{ switch((unsigned char) fmtp[0])
{ case '#':
fmtp++
; bufp = NULL
; continue
;
case 'n':
if (bufp != bufz)
goto DONE
; fmtp++
; bufp = NULL
; continue
;
case '%':
if (bufp[0] != '%') goto DONE
; bufp++
; break
;
case 't':
if (bufp[0] != '\t') goto DONE
; bufp++
; break
;
case 's':
if (bufp[0] != ' ') goto DONE
; bufp++
; break
;
case '.':
bufp++
; break
;
case 'b':
while (bufp < bufz && izblank(bufp[0]))
bufp++
; break
;
case 'C': case 'X':
{ int n_scanned = 0
; if (sscanf(bufp, "%u%n", &rec->count, &n_scanned) < 1)
goto DONE
; bufp += n_scanned
; }
break
;
case 'G':
while (bufp < bufz && !izblank((unsigned char) bufp[0]))
bufp++
; rec->discard_n = cpytofield(rec->discard, curp, bufp -curp)
; break
;
case 'F':
while (bufp < bufz && '\t' != (unsigned char) bufp[0])
bufp++
; rec->discard_n = cpytofield(rec->discard, curp, bufp -curp)
; break
;
case 'H':
while (bufp < bufz && (unsigned char) bufp[0] != (unsigned char) fmtp[1])
bufp++
; if (bufp == bufz)
goto DONE
; rec->discard_n = cpytofield(rec->discard, curp, bufp -curp)
; fmtp++
; bufp++
; break
;
case 'a':
rec->annot_n = cpytofield(rec->annot, rec->discard, rec->discard_n)
; break
;
case 'A':
while (izblank((unsigned char) curp[0]))
curp++
; bufp = curp
; while (bufp < bufz)
bufp++
; rec->annot_n = cpytofield(rec->annot, curp, bufp -curp)
; break
;
case 'I':
while (bufp < bufz && !izblank((unsigned char) bufp[0]))
bufp++
; rec->id_n = cpytofield(rec->id, curp, bufp -curp)
; break
;
case 'Q':
while (bufp < bufz && !izblank((unsigned char) bufp[0]))
bufp++
; rec->q_n = cpytofield(rec->q, curp, bufp -curp)
; break
;
case 'R':
while (bufp < bufz && isalpha((unsigned char) bufp[0]))
bufp++
; rec->seq_n = cpytofield(rec->seq, curp, bufp -curp)
; validate_sequence(rec)
; break
;
default:
goto DONE
; }
esc = 0
; }
else
{ if (fmtp[0] == '%')
esc = 1
; else if (bufp[0] != fmtp[0])
goto DONE
; else
bufp++
; }
fmtp++
; }
DONE :
if (truncated)
argh("reaper", "line too long")
, rlstat |= RL_ERROR
; if (n_lines && (fmtp[0] || (bufp && bufp - buf != n_received)))
{ argh("reaper", "parse error (remaining format string %s, buffer [%s])", fmtp, bufp)
; return RR_ERROR
; }
if (rlstat & RL_DONE)
return RR_DONE
; if (rlstat & RL_ERROR)
return RR_ERROR
; if (rlstat & RL_NOMEM) /* fixme not checked yet */
return RR_NOMEM
; rec->rs.N++
; return RR_OK
; }
/*
/stringoffsets/
=========================================================
alnspace
alignment space:
0 1 2 3 4 5 6 7 8 9 10 coordinates
G G G A T C C T A G read
C T A G G C C adaptor
left_start : 7
lint_offset : left_start - 1 == 6 == clean_n
=========================================================
string space:
0 1 2 3 4 5 6 7 8 9 10 coordinates
G G G A N C N T N G read
^ N offset
N offset is 4, we return 5, so that
lint_offset == NNN_offset - 1 == 4 == clean_n
========================================================= */
int get_nnn_offset
( const char* s
, unsigned n
, unsigned count
, unsigned outof
)
{ unsigned hist[SW_ALN_MAXSIZE] = { 0 }
; if (!outof || count > outof)
return -1
; int i, nn = 0
; for (i=0;i<n;i++)
{ unsigned char a = s[i]
; unsigned char e = hist[i % outof]
; hist[i % outof] = a
; nn += (a == 'N') - (e == 'N')
; if (nn >= count)
{ while (nn > 0) /* now check which ones are not N */
{ if (hist[i-- % outof] == 'N')
nn--
; }
return i+2 /* noteme, alignment space coordinate
* i+1 is the string-space NNNN offset,
* but we need to return offset in aligment space (see above)
*/
; }
}
return -1
; }
/* same return semantics as get_nnn_offset */
void get_dust_info
( const char* s
, unsigned n
, int* maxp /* dust score for tail */
, int* maxidp /* offset where this is attained */
, int* dominant_base /* dominant base */
, int* dominant_base_count /* and its count */
)
{ int max = 0, maxid =0, i
; unsigned count[6] = { 0 }
; dustscore_tail(s, n, NULL, &max, &maxid)
; maxp[0] = max
; maxidp[0] = maxid
; dominant_base[0] = 4
; dominant_base_count[0] = 0
; for (i=maxid; i<n; i++)
count[themap[(unsigned char) s[i]]]++
; for (i=0;i<4;i++)
if (count[i] > dominant_base_count[0])
dominant_base[0] = i
, dominant_base_count[0] = count[i]
; }
/* same return semantics as get_nnn_offset */
int get_dust_offset
( const char* s
, unsigned n
, unsigned threshold
, unsigned bases
)
{ int max = 0, maxid =0, found = 0
; dustscore_tail(s, n, NULL, &max, &maxid)
; found = threshold && max >= threshold && maxid+6 <= n
; if (found && bases)
{ int i
; unsigned count[6] = { 0 }
; for (i=maxid; i<n; i++) /* fixme: i=0 should that be i=maxid ? */
count[themap[(unsigned char) s[i]]]++
; for (i=0;i<4;i++)
{ if (count[i] * 2 > (n - maxid) && (bases & (1 << i)))
break
; }
found = i < 4
; }
return found ? (maxid+1) : -1
; }
/* same return semantics as get_nnn_offset */
int get_polya_offset
( const char* s
, unsigned n
, int minlen
)
{ int i = n /* Note: should work for q_n == 0 */
; int polya = 0
; while (--i >= 0 && s[i] == 'A')
polya++
; return minlen && polya >= minlen ? n + 1 - polya : -1
; }
int cmp_int
( const void* a
, const void* b
)
{ return ((int*)a)[0] - ((int*)b)[0]
; }
int get_qqq_offset
( struct record* rec
, int cut /* cut when median drops below this */
, unsigned winlen
, unsigned winofs
, unsigned readofs
)
{ unsigned char hist[256] = { 0 }
; int window[255] /* signed so that we can use a-b for comparison */
; unsigned winmid
; int i = 0, n_lt = 0, n_gt = 0
; int median = 0
; int lastqual = 0
; if (!winlen || readofs + winlen > rec->q_n)
return -1
; lastqual = rec->q[rec->q_n-1]
; winmid = (winlen+1) / 2 /* e.g. 5 -> 3; n_smaller + n_equal must be >= winmid
* and also n_larger + n_equal must be >= winmid
*/
; for (i=readofs;i<winlen+readofs; i++)
window[i % winlen] = rec->q[i]
; qsort(window, winlen, sizeof window[0], cmp_int)
; median = window[winmid-1]
; for (i=readofs;i<winlen+readofs; i++)
{ unsigned i_out = i % winlen
; window[i_out] = rec->q[i]
; hist[(unsigned char) window[i_out]]++
; n_lt += window[i_out] < median
; n_gt += window[i_out] > median
; }
i = readofs + winlen - 1 /* necessary for median >= cut logic */
; while (median >= cut && ++i < rec->q_n + winlen)
{ unsigned i_out = i % winlen
; unsigned rgt = i < rec->q_n ? rec->q[i] : lastqual /* shift this value in */
; unsigned lft = window[i_out] /* shift this value out */
; window[i_out] = rgt
; hist[rgt]++
; hist[lft]--
; if (rgt < median) n_lt++
; if (rgt > median) n_gt++
; if (lft < median) n_lt--
; if (lft > median) n_gt--
; while (!hist[median] || n_lt + hist[median] < winmid)
{ n_lt += hist[median]
; n_gt -= hist[median+1]
; median++
; }
while (!hist[median] || n_gt + hist[median] < winmid)
{ n_gt += hist[median]
; n_lt -= hist[median-1]
; median--
; }
;if(0)fprintf(stderr, " %2d-%2d-%d", ji median, ji rgt, ji i);
}
;if(0)fputc('\n', stderr);
{ int j = (i+1+winofs)-winlen
; if (j > rec->q_n)
j = rec->q_n
; if (i < (rec->q_n + winlen))
return j
; }
return -1
; }
int get_bbb_offset
( struct record* rec
)
{ int i = rec->q_n /* Note: should work for q_n == 0 */
; while (--i >= 0)
{ if (rec->q[i] != 'B')
break
; }
return i+1 == rec->q_n ? -1 : i+2 /* same return semantics as get_nnn_offset */
; }
double string_complexity
( const char* a
, unsigned n
)
{ int i, n_pairs = 0
; if (!n)
return 1.0
; for (i=0;i+1<n;i++)
n_pairs += a[i] == a[i+1]
; return (1.0 - n_pairs * 1.0 / n)
; }
#if 0
/* fixme: n_stretch in callers often computed using SW_ALN_MAXSIZE, but trimming changes this */
/* returns number of matches if parameters satisfied, 0 otherwise */
int alignment_edit_ok
( const unsigned char* aln
, unsigned aln_length
, unsigned n_stretch
, unsigned n_edit_max
, unsigned n_gap_max
)
{ int i, n_edit = 0, n_gap = 0, n_subst = 0, n_match = 0, ok = 0
; if (!n_stretch)
return 0
; unsigned hist[SW_ALN_MAXSIZE] = { 0 }
;if(0)fprintf(stderr, "length %d n_stretch %d e %d\n", (int) aln_length , (int) n_stretch , (int) n_edit_max)
; for (i=0;i<aln_length;i++)
{ unsigned char a = aln[i]
; unsigned char e = hist[i % n_stretch]
; hist[i % n_stretch] = a
; n_subst += (a == 'x') - (e == 'x')
; n_gap += (a == '-') - (e == '-')
, n_edit = n_subst + n_gap
; n_match += a == '|'
; if ( n_edit <= n_edit_max
&& i+1 + n_edit_max >= n_stretch + n_edit
&& n_gap <= n_gap_max
)
ok = 1
; }
return ok ? n_match : 0
; }
#endif
/* the prefix match is complicated.
* We must be able to find a good match at the end
* that's not beaten by a longer shoddy match at start
* for which the suffix complexity criterion fails. This
* probably, ideally, requires using the last-row max_ej cell.
* The max_ej cell will not, however, find prefix matches
* followed by low complexity reads.
*/
int sw_match_end
( struct sw_alninfo* ai
, struct match_account* ma
, struct sw_param* swp
, unsigned prefix_perfect_minlen
, struct match_requirement* mr_prefix
)
{ unsigned n_gap = ai->n_insl + ai->n_insr
; unsigned n_edit = n_gap + ai->n_subst
; unsigned prefix_length = ai->rgt_end + 1 - ai->rgt_start
; ma->suffix_complexity = 0.0
/* This code uses the globally highest scoring alignment,
* and applies the prefix match requirement to the entire
* alignment, not just a subpart.
* This is so that a case such as
GATAGATCGAAACCATCCTCTGCTAGTTTTC-GTATG : [16,36]
||x|x|||-x||x|||-|xx|| : score 56
TCGTATGC-CGTCTTCTGCTTGT : [1,21] 0/0 recno=1
* is not considered a match.
*/
; unsigned prefix_match
= ( (!mr_prefix->mr_offset || (ai->rgt_start <= mr_prefix->mr_offset))
&& prefix_length >= mr_prefix->mr_minlen
&& n_edit <= mr_prefix->mr_maxedit
&& n_gap <= mr_prefix->mr_maxgap
)
/* now we just inspect the end. */
; if (!prefix_match && prefix_perfect_minlen)
{ int ofs = (ai->ni-1) * ai->nj /* this sits on the left column of zeros, last entry */
; unsigned j, bestj = 0
; for (j=prefix_perfect_minlen;j<ai->nj;j++)
{ if (ai->data[ofs+j] == j * swp->gain_match)
bestj = j
; }
ma->n_perfect = bestj
; return bestj ? 1 : 0
; }
/* NOTE. The prefix match may fall just short of the read,
* for example miss the last one or two bases. The suffix
* complexity will still be small, as a result of
* prefix_length (the matching part of the adaptor)
* being part of the formula
*/
if (prefix_match)
{ if (ai->lft_end + 1 < ai->ni)
{ double left_trail = ai->ni - ai->lft_end - 1
; double c = string_complexity(ai->left+ai->lft_end+1, ai->ni - ai->lft_end-1)
; ma->suffix_complexity = c * (left_trail) / (1.0 * prefix_length + left_trail)
; }
return 1
; }
return 0
; }
/* barcode code */
int sw_match_bc3p
( struct sw_alninfo* ai
, struct match_requirement* mr_global
, struct match_requirement* mr_prefix
, struct match_requirement* mr_barcode
, int sinsert_n /* number of bases before the barcode, not yet used. */
, int barcode_n /* length of barcode, not yet used. */
, int* n_match
)
{ int n_length = mr_global->mr_minlen
; int n_edit_max = mr_global->mr_maxedit
; int n_gap_max = mr_global->mr_maxgap
; n_match[0] = 0
;
/* /barcodehardcoded/
* note: 3p-sinsert+barcode start hardcoded to be in pos 1 or 2
* (beware of alignment space, which uses offset 1)
* Now we have to be careful wrt to the barcode match requirements;
* How do we locate the right spot in the alignment?
*/
if (ai->rgt_start != 1 && ai->rgt_start != 2)
{ if(0)fprintf(stderr, "bail-out 2 (%d)\n", (int) ai->rgt_start)
; return 0
; }
/* reduce global range alignment length parameter if match is at end of range */
if (ai->ni - ai->lft_start < n_length && mr_prefix->mr_minlen)
{ n_length = ai->ni - ai->lft_start
; if (n_length < mr_prefix->mr_minlen)
return 0
; n_edit_max = mr_prefix->mr_maxedit
; n_gap_max = mr_prefix->mr_maxgap
; }
return
( ( !mr_barcode->mr_minlen /* check barcode range alignment if specied */
|| sw_alignment_edit_ok
( ai->aln_aln + ai->aln_ofs
/* fixme, better control over barcode */
, mr_barcode->mr_minlen /* fixed length search: can this overrun the alignment? document! */
, mr_barcode->mr_minlen
, mr_barcode->mr_maxedit
, mr_barcode->mr_maxgap
)
)
&&
/* second argument, alignment length: this is not
* strictly correct, but the remainder of the
* alignment is zero-filled. Note: our search
* space (right_limit in sw_fill2) has been
* restricted in do_barcode_3p.
*/
#define WOTCHA 0
/* check global range alignment */
( n_match[0]
= sw_alignment_edit_ok
( ai->aln_aln + ai->aln_ofs
, ai->aln_end - ai->aln_ofs
, n_length
, n_edit_max
, n_gap_max
) )
)
; }
/* aln_ofs == SW_ALN_MAXSIZE + 1 */
int sw_match_any
( struct sw_alninfo* ai
, struct match_requirement* mr
)
{
return
( !mr->mr_offset
|| (mr->mr_offset > 0 && ai->rgt_start <= mr->mr_offset)
|| (mr->mr_offset < 0 && (ai->nj - ai->rgt_end) <= -mr->mr_offset)
)
&& sw_alignment_edit_ok
( ai->aln_aln + ai->aln_ofs
, ai->aln_end - ai->aln_ofs
, mr->mr_minlen
, mr->mr_maxedit
, mr->mr_maxgap
)
; }
int sw_printaccount
( const struct match_account* ma
, char* buf
, unsigned bufsize
)
{ return
snprintf
( buf
, bufsize
, "mt=%c,sc=%.2f,ht=%d,nn=%d,bb=%d,aa=%d,qq=%d"
, (int) ma->match_type
, ma->suffix_complexity
, (int) ma->n_perfect
, (int) ma->nnn_offset
, (int) ma->bbb_offset
, (int) ma->aaa_offset
, (int) ma->qqq_offset
)
; }
int sw_printaln3
( const struct sw_alninfo* ai
, char* buf
, unsigned bufsize
)
{ unsigned o = ai->aln_ofs
; return
snprintf
( buf
, bufsize
, "ls=%d,le=%d,rs=%d,re=%d,score=%d,left=%s,aln=%s,right=%s"
, (int) ai->lft_start
, (int) ai->lft_end
, (int) ai->rgt_start
, (int) ai->rgt_end
, (int) ai->data[ai->max_ij]
, ai->aln_lft+o
, ai->aln_aln+o
, ai->aln_rgt+o
)
; }
void dump_read
( int cleaned
, void* fpo
, const struct record* rec
, const struct sw_alninfo* ai
, const struct match_account* ma
, struct param* pam
)
#define THEBUFSIZE 1023
{ char buf[THEBUFSIZE+1]
; char buf2[THEBUFSIZE+1]
#if WE_USE_ZLIB
; int zippit = pam->zippit
#endif
; const char* format = cleaned ? pam->format_clean : pam->format_lint
; const char* f = format
; const char* z = format + strlen(format)
; char dust [MAXFIELDSIZE+1]
; int escape = 0, tnt
; int dustidx = 0, dustscore = 0
; while (f < z)
{ unsigned c = (unsigned char) f[0]
; const char* p = buf
; int n = 0
; if (!escape)
{ if (c == '%')
escape = 1
; else
buf[0] = c
, n = 1
; }
else
{ int cleanlen = cleaned ? rec->clean_n : rec->seq_n
; switch(c)
{ case 'R': p = rec->seq; n = rec->seq_n;
break;
; case 'C': case 'E': p = rec->seq; n = rec->clean_n;
if (c == 'E' && n == 0) { p = "N"; n = 1; }
break;
; case 'V': revcompl(rec->seq, rec->clean_n, buf); p = buf; n = rec->clean_n;
break;
; case 'Z': { int themax = rec->seq_n;
if (pam->length_co > themax) themax = pam->length_co;
n = snprintf(buf, THEBUFSIZE+1, "%.*s", (int) rec->clean_n, rec->seq);
while (n < themax && n < THEBUFSIZE)
buf[n++] = 'N';
buf[n] = '\0';
break;
}
; case 'I': p = rec->id; n = rec->id_n; break;
; case 'L': n = snprintf(buf, THEBUFSIZE+1, "%d", (int) rec->clean_n); break;
; case 'X': n = snprintf(buf, THEBUFSIZE+1, "%u", (unsigned) rec->count); break;
; case 'Q': p = rec->q; n = cleanlen; break;
; case 'Y': p = rec->q; n = rec->q_n; break;
; case 'U': { int score, score_offset, base, base_count
; int seqlen = cleanlen
; get_dust_info(rec->seq, seqlen, &score, &score_offset, &base, &base_count)
; n = snprintf(buf, THEBUFSIZE+1, "score=%d,offset=%d,len=%d,base=%c,count=%d",
score, score_offset+1, seqlen - score_offset, dna[base], base_count)
; }
; break
; case 'T': tnt = trintscore(rec->seq, cleanlen)
; n = snprintf(buf, THEBUFSIZE+1, "%d", tnt)
; break
; case 'D': dustscore_tail(rec->seq, rec->clean_n, dust, &dustscore, &dustidx)
; p = dust; n = rec->clean_n; break
; case '_': if (!dustscore)
dustscore_tail(rec->seq, rec->seq_n, NULL, &dustscore, &dustidx)
; n = snprintf(buf, THEBUFSIZE+1, "%d:%d", dustidx+1, dustscore);
break
; case 'M': p = rec->out_message; n = strlen(p); break
; case 'i': case 'J': n = snprintf(buf, THEBUFSIZE+1, "%d", (int) rec->rs.N); break;
; case 'f': n = snprintf(buf, THEBUFSIZE+1, "%d", (int) (4 * (rec->rs.N-1) + 2)); break;
; case '3': if (ai) n = sw_printaln3(ai, buf, THEBUFSIZE+1); break
; case 'A': p = rec->annot; n = rec->annot_n; break
; case '?': n = get_mdta(buf, buf2, rec->seq, rec->seq_n, pam); break
; case '=': if (ma) n = sw_printaccount(ma, buf, THEBUFSIZE+1); break
; case 'n':
case 't':
case 's':
case '%': buf[0] = c == 'n' ? '\n' : c == 't' ? '\t' : c == 's' ? ' ' : '%'
; n = 1
; break
; case 'q': { int a = (unsigned char) f[1];
int themax = cleanlen;
if (pam->length_co > themax) themax = pam->length_co;
f++;
n = snprintf(buf, THEBUFSIZE+1, "%.*s", (int) rec->clean_n, rec->q);
while (n < themax && n < THEBUFSIZE)
buf[n++] = a;
buf[n] = '\0';
break;
}
; }
escape = 0
; }
if (n > 0)
{
#if WE_USE_ZLIB
if (zippit) gzwrite(fpo, p, n)
; else fwrite(p, n, 1, fpo)
#else
fwrite(p, n, 1, fpo)
#endif
; }
f++
; }
}
#undef THEBUFSIZE
/* low complexity */
int loco
( struct record* rec
, struct param* pam
, double loco_check
)
{ if (string_complexity(rec->seq, rec->seq_n) < loco_check)
{ rec->out_message = "loco"
; dump_read(0, pam->fplint, rec, NULL, NULL, pam)
; return 1
; }
return 0
; }
/* hierverder segfault if buckid == buck_n or if maxfieldsize == MAXFIELDSIZE.
should be plenty, no?
*/
void qc_tally_in
( struct record* rec
, struct param* pam
, int buckid
, int N
)
{ int j
; for (j=0;j<N;j++)
{ switch(rec->seq[j])
{ case 'A': pam->buck[buckid].qc_cpsin.A[j]++; break
; case 'C': pam->buck[buckid].qc_cpsin.C[j]++; break
; case 'G': pam->buck[buckid].qc_cpsin.G[j]++; break
; case 'T': pam->buck[buckid].qc_cpsin.T[j]++; break
; case 'N': pam->buck[buckid].qc_cpsin.N[j]++; break
; }
if (rec->q_n)
{ int kk = MAXQUALITY * j + ((unsigned char) rec->q[j])
;if(0)fprintf(stderr, "testing2 recid=%d buckid=%d qid=%d nt=%d val=%d rootp=%p kk=%d seq=%s\n"
, (int) rec->rs.N
, buckid
, (int) (7 & (unsigned char) rec->q[j])
, (int) j
, 0
, (void*) (pam->buck[buckid].qc_bcq)
, kk
, rec->seq
)
; pam->buck[buckid].qc_bcq[kk] += 1 /* one block is one read-offset */
/* [ nt0 size MAXQUALITY ] [ nt1 size MAXQUALITY] */
; }
}
if (N > pam->buck[buckid].global_in_max_size)
pam->buck[buckid].global_in_max_size = N
; }
void qc_tally_out
( struct record* rec
, struct param* pam
, int buckid
, int N
)
{ int j
; pam->buck[buckid].qc_lenout[N]++
; for (j=0;j<N;j++)
{ switch(rec->seq[j])
{ case 'A': pam->buck[buckid].qc_cpsout.A[j]++; break
; case 'C': pam->buck[buckid].qc_cpsout.C[j]++; break
; case 'G': pam->buck[buckid].qc_cpsout.G[j]++; break
; case 'T': pam->buck[buckid].qc_cpsout.T[j]++; break
; case 'N': pam->buck[buckid].qc_cpsout.N[j]++; break
; }
}
if (N > pam->buck[buckid].global_out_max_size)
pam->buck[buckid].global_out_max_size = N
; }
void qc_brief
( struct param *pam
, struct record *rec
, enum program_mode mode
)
{ argh
( "reaper"
, "check %d errors, %d reads truncated, %d clean, %d lint, %d total"
, (int) rec->rs.N_err
, (int) rec->rs.N_truncated
, (int) rec->rs.N_clean
, (int) rec->rs.N_lint
, (int) rec->rs.N
)
; if (rec->rs.N_alien)
argh
( "reaper"
, "found %d characters outside [ACGTUNX], average %.1f per read"
, rec->rs.N_alien
, rec->rs.N_alien * 1.0 / rec->rs.N
)
; if (pam->buck_n > 1)
{ int i, j
; for (i=0;i<pam->buck_n;i++)
{ dim sum = 0
; const char* bc = pam->buck[i].barcode5p
; if (!bc)
bc = pam->buck[i].barcode3p
; for (j=0; j<MAXFIELDSIZE; j++)
sum += pam->buck[i].qc_lenout[j]
; argh("reaper", "barcode [%s] has %lu clean reads", bc, (long unsigned) sum)
; }
}
}
/* this is the quality control report function. mat-happy! */
void qc_report
( struct param* pam
, struct record* rec
, enum program_mode mode
)
{ int i, j, buckid
; for (buckid=0;buckid<=pam->buck_n;buckid++)
{ struct bucket* buck = pam->buck+buckid
; const char* stage2 = "clean"
; const char* thetag = buck->filetag
; if (!thetag)
thetag = buck->barcode3p
; if (!thetag) /* hack, but OK */
thetag = buck->barcode5p
; if (!thetag) /* hack hack hack */
thetag = "lane"
/* todo: qc_cpsin and qc_cpsout need to keep track of maximum read lengths */
/* qc hack again */
; if (buckid == pam->buck_n && (mode == CASE_3P_BC || mode == CASE_5P_BC))
thetag = "total"
, stage2 = "miss"
; char* fn_pre_nt = stringle("%s.%s.report.input.nt", pam->basename, thetag)
; char* fn_pre_q = stringle("%s.%s.report.input.q", pam->basename, thetag)
; char* fn_post_nt = stringle("%s.%s.report.%s.nt", pam->basename, thetag, stage2)
; char* fn_post_q = stringle("%s.%s.report.%s.q", pam->basename, thetag, stage2)
; char* fn_post_len= stringle("%s.%s.report.%s.len", pam->basename, thetag, stage2)
; FILE* fp_pre_nt, *fp_pre_q
,* fp_post_nt, *fp_post_len
; dim *cpsout[5]
= { buck->qc_cpsout.A
, buck->qc_cpsout.C
, buck->qc_cpsout.G
, buck->qc_cpsout.T
, buck->qc_cpsout.N
}
, *cpsin[5]
= { buck->qc_cpsin.A
, buck->qc_cpsin.C
, buck->qc_cpsin.G
, buck->qc_cpsin.T
, buck->qc_cpsin.N
}
; if ((fp_pre_nt = myfopen(fn_pre_nt, "w", 0)))
{ fprintf(fp_pre_nt, "pos\tA\tC\tG\tT\tN\n")
; for (j=0;j<buck->global_in_max_size;j++)
fprintf
( fp_pre_nt
, "%d\t%d\t%d\t%d\t%d\t%d\n"
, (int) (j+1)
, (int) cpsin[0][j]
, (int) cpsin[1][j]
, (int) cpsin[2][j]
, (int) cpsin[3][j]
, (int) cpsin[4][j]
)
; myfzclose(fp_pre_nt, 0)
; }
if ((fp_post_nt = myfopen(fn_post_nt, "w", 0)))
{ fprintf(fp_post_nt, "pos\tA\tC\tG\tT\tN\n")
; for (j=0;j<buck->global_out_max_size;j++)
fprintf
( fp_post_nt
, "%d\t%d\t%d\t%d\t%d\t%d\n"
, (int) (j+1)
, (int) cpsout[0][j]
, (int) cpsout[1][j]
, (int) cpsout[2][j]
, (int) cpsout[3][j]
, (int) cpsout[4][j]
)
; myfzclose(fp_post_nt, 0)
; }
if ((fp_post_len = myfopen(fn_post_len, "w", 0)))
{ fprintf(fp_post_len, "length\tcount\n")
; for (i=0;i<=buck->global_out_max_size;i++)
fprintf
( fp_post_len
, "%d\t%d\n"
, (int) i
, (int) pam->buck[buckid].qc_lenout[i]
)
; myfzclose(fp_post_len, 0)
; }
if ((fp_pre_q = myfopen(fn_pre_q, "w", 0)))
{ fprintf(fp_pre_q, "pos\tq0\tq10\tq25\tq50\tq75\tq90\tq100\n")
; for (i=0;i<pam->buck[buckid].global_in_max_size;i++)
{ dim total = 0, total2 = 0, j /* fixme overflow */
; dim q0=0, q10=0, q25=0, q50=0, q75=0, q90=0, q100=0
; for (j=1;j<MAXQUALITY;j++) /* offset 1 because of q0 check */
total += buck->qc_bcq[i * MAXQUALITY + j]
; for (j=1;j<MAXQUALITY;j++)
{ int kk = i * MAXQUALITY + j
; total2 += buck->qc_bcq[kk]
; if (!q0 && total * 0.0 < total2)
q0 = j
; if (!q10 && total * 0.10 <= total2)
q10 = j
; if (!q25 && total * 0.25 <= total2)
q25 = j
; if (!q50 && total * 0.50 <= total2)
q50 = j
; if (!q75 && total * 0.75 <= total2)
q75 = j
; if (!q90 && total * 0.90 <= total2)
q90 = j
; if (!q100 && total * 0.9999 <= total2)
q100 = j
; }
fprintf
( fp_pre_q, "%d\t%d\t%d\t%d\t%d\t%d\t%d\t%d\n"
, (int) (i+1)
, (int) q0
, (int) q10
, (int) q25
, (int) q50
, (int) q75
, (int) q90
, (int) q100
)
; }
myfzclose(fp_pre_q, 0)
; }
free(fn_pre_nt)
; free(fn_pre_q)
; free(fn_post_nt)
; free(fn_post_q)
; free(fn_post_len)
; }
{ char* fn_post_tally= stringle("%s.sumstat", pam->basename)
; FILE* fp_post_tally
; if ((fp_post_tally = myfopen(fn_post_tally, "w", 0)))
{ fprintf(fp_post_tally, "discarded_no_barcode_match=%u\n", rec->rs.N_nomatch)
; fprintf(fp_post_tally, "discarded_tabu_contamination=%u\n", rec->rs.N_tabu)
; fprintf(fp_post_tally, "discarded_no_5p_sinsert=%u\n", rec->rs.N_5p_nosinsert)
; fprintf(fp_post_tally, "discarded_no_anchor=%u\n", rec->rs.N_anchor)
; fprintf(fp_post_tally, "discarded_NNN=%u\n", rec->rs.N_nn)
; fprintf(fp_post_tally, "discarded_BBB=%u\n", rec->rs.N_bb)
; fprintf(fp_post_tally, "discarded_AAA=%u\n", rec->rs.N_aa)
; fprintf(fp_post_tally, "discarded_QQQ=%u\n", rec->rs.N_qq)
; fprintf(fp_post_tally, "discarded_tri=%u\n", rec->rs.N_tri)
; fprintf(fp_post_tally, "discarded_length_cutoff=%u\n", rec->rs.N_lenlo)
; fprintf(fp_post_tally, "discarded_barcode_conflict=%u\n", rec->rs.N_cfl)
; fprintf(fp_post_tally, "discarded_low_complexity=%u\n", rec->rs.N_loco)
; fprintf(fp_post_tally, "no_message=%u\n", rec->rs.N_nomsg)
; fprintf(fp_post_tally, "total_discarded=%u\n", rec->rs.N_lint)
; fprintf(fp_post_tally, "total_error=%u\n", rec->rs.N_err)
; fprintf(fp_post_tally, "total_accepted=%u\n", rec->rs.N_clean)
; fprintf(fp_post_tally, "total_input=%u\n", rec->rs.N)
; fprintf(fp_post_tally, "conflicts_resolved=%u\n", rec->rs.N_cflr)
; fprintf(fp_post_tally, "bases_removed_by_quality=%u\n", rec->rs.B_QQ)
; fprintf(fp_post_tally, "bases_removed_by_qualityb=%u\n", rec->rs.B_BB)
; fprintf(fp_post_tally, "bases_removed_by_tritail=%u\n", rec->rs.B_tri)
; fprintf(fp_post_tally, "bases_removed_by_nnn=%u\n", rec->rs.B_NN)
; fprintf(fp_post_tally, "bases_removed_by_polya=%u\n", rec->rs.B_AA)
; myfzclose(fp_post_tally, 0)
; free(fn_post_tally)
; }
}
}
int buckets_prepare_qc
( struct bucket* buck
, int buck_n
, enum program_mode mode
)
{ int i
; for (i=0;i<=buck_n;i++)
{ buck[i].cache_lint_offset = 0
; buck[i].cache_score = 0
; buck[i].cache_accept = 0
; memset(buck[i].qc_cpsin.A, 0, sizeof buck[i].qc_cpsin.A)
; memset(buck[i].qc_cpsin.C, 0, sizeof buck[i].qc_cpsin.C)
; memset(buck[i].qc_cpsin.G, 0, sizeof buck[i].qc_cpsin.G)
; memset(buck[i].qc_cpsin.T, 0, sizeof buck[i].qc_cpsin.T)
; memset(buck[i].qc_cpsin.N, 0, sizeof buck[i].qc_cpsin.N)
; memset(buck[i].qc_cpsout.A, 0, sizeof buck[i].qc_cpsout.A)
; memset(buck[i].qc_cpsout.C, 0, sizeof buck[i].qc_cpsout.C)
; memset(buck[i].qc_cpsout.G, 0, sizeof buck[i].qc_cpsout.G)
; memset(buck[i].qc_cpsout.T, 0, sizeof buck[i].qc_cpsout.T)
; memset(buck[i].qc_cpsout.N, 0, sizeof buck[i].qc_cpsout.N)
; memset(buck[i].qc_lenout, 0, sizeof buck[i].qc_lenout)
/* fixme memrelease */
; // if (!(buck[i].qc_bcq = calloc(1 + (MAXQUALITY * MAXFIELDSIZE), sizeof(dim))))
; if (!(buck[i].qc_bcq = calloc((MAXQUALITY * MAXFIELDSIZE) , sizeof(dim))))
X_FINISHED("cannot allocate memory for quality tracking")
; buck[i].global_in_max_size = 0
; buck[i].global_out_max_size = 0
; buck[i].barcode_start = 0
; buck[i].barcode_end = 0
; }
/* fixme very ugly; init code only for last "special" bucket */
/* rip and refactor at some point */
buck[buck_n].barcode3p = NULL
; buck[buck_n].adaptor3p = "the-3-prime-adaptor"
; buck[buck_n].sinsert3p = NULL
; buck[buck_n].barcode5p = NULL
; buck[buck_n].sinsert5p = NULL
; buck[buck_n].cat3p = NULL
; buck[buck_n].filetag = NULL
; return 0
; }
void data_release
( struct param* pam
)
{ int i
; if (pam->buck)
{ for (i=0;i<=pam->buck_n;i++)
{ struct bucket* buck = pam->buck+i
; struct atabu* boo = &buck->tabu2, *boop = NULL /* boop'revious */
; int n = 0
; while (boo) /* hierverder: corrupt if meta file cannot be opened! */
{ if (boo->tabuseq)
free(boo->tabuseq)
; boop = boo->next
; if (n++)
free(boo)
; boo = boop
; }
if (buck->qc_bcq) free(buck->qc_bcq)
; if (buck->fname) free(buck->fname)
; }
free(pam->buck)
; }
pam->buck = NULL
; pam->buck_n = -1
; table_release(&pam->metadata)
; }
int barcodes_io_open
( struct bucket* buck
, const char* basename
, int buck_n
, int zippit
)
{ int i
; for (i=0;i<buck_n;i++)
{ const char* thetag = buck[i].filetag
; if (!thetag)
thetag = buck[i].barcode3p
; if (!thetag)
thetag = buck[i].barcode5p
; buck[i].fname = stringle("%s.%s.%s", basename, thetag, zippit ? "clean.gz" : "clean")
; if (!(buck[i].fp = myfopen(buck[i].fname, ZWMODE, zippit)))
{ argh("reaper", "cannot open file %s", buck[i].fname)
; return 1
; }
}
return 0
; }
int barcodes_io_close
( struct bucket* buck
, int buck_n
, int zippit
)
{ int i
; for (i=0;i<buck_n;i++)
myfzclose(buck[i].fp, zippit)
; return 0
; }
/* fixme, could have seqp and qp pointers */
void rec_remove_prefix
( struct record* rec
, int rm
)
{ if (rm > rec->seq_n)
rm = rec->seq_n
; memmove(rec->seq, rec->seq+rm, rec->seq_n - rm)
; rec->seq_n -= rm
; rec->seq[rec->seq_n] = '\0'
; if (rec->q_n)
memmove(rec->q, rec->q+rm, rec->q_n - rm)
, rec->q_n -= rm
, rec->q[rec->q_n] = '\0'
; }
int column_require
( int i
, const char* str
)
{ if (i < 0)
X_FINISHED("need column %s", str)
; return 0
; }
/* fixme: seem to have multiple places with barcode instantiation.
* deduplicate.
*/
int adaptors_set_case_3p
( struct bucket* buck
, int* buck_n
, struct param* pam
)
{ buck[0].barcode5p = NULL
; buck[0].barcode3p = NULL
; buck[0].sinsert3p = NULL
; buck[0].cat3p = NULL
; buck[0].sinsert5p = pam->sinsert_5p
; buck[0].adaptor3p = pam->adaptor_3p
; buck[0].tabu2 = pam->the_tabu2
; if (g_mode == CASE_NOBC && !pam->adaptor_3p)
X_FINISHED("-geom 3p need a 3p adaptor")
; if
( g_restrict > 0
&& strlen(pam->adaptor_3p) > g_restrict
/* fixme: old code had
&& strlen(pam->adaptor_3p) < g_restrict
and led to different results sometimes.
*/
)
((char*) buck[0].adaptor3p)[g_restrict] = '\0'
;fprintf(stderr, "adapter %s\n", buck[0].adaptor3p)
; buck_n[0] = 1
; return 0
; }
/* ugly, designtabu */
void parse_tabu
( struct atabu* boo
, const char* s
, int extent
)
{ struct match_requirement mr_tabu = { 16, 2, 1, 0 }
; const char* tok1 = NULL
; do
{ char* s2 = NULL
; tok1 = strchr(s, ',')
; s2 = stringle("%.*s", (int) (tok1 ? (tok1 - s) : strlen(s)), s)
; boo->tabuseq = s2
; boo->tabulen = strlen(s2)
; boo->mr_tabu = mr_tabu
; boo->next = NULL
; if (extent && boo->tabulen > extent)
boo->tabulen = extent
, boo->tabuseq[extent] = '\0'
; if (tok1)
{ boo->next = myalloc(sizeof boo[0])
; boo = boo->next
; s = tok1 + 1
; }
}
while(tok1)
; return
; }
int buckets_alloc
( struct param* pam
, int n
)
{ if (n > 64)
argh("reaper", "that is a lot of barcodes, that is. %d of'em", n)
; if (!(pam->buck = calloc((n+1), sizeof pam->buck[0])))
X_FINISHED("unable to allocate data for %d barcodes", n)
; pam->buck_n = n
; return 0
; }
int config_read
( const char* fn
, struct param* pam
, enum program_mode mode
, int extent
)
{ int id_barcode , doc_barcode = 0
; int id_3p_ad , doc_3p_ad = 1
; int id_3p_si , doc_3p_si = 2
; int id_5p_si , doc_5p_si = 3
; int id_tabu , doc_tabu = 4
; int id_filetag , doc_filetag = 5
; int n_columns_found
; int n_barcodes = 0
; int barcodes_length = 0
; struct table* thetable = &pam->metadata
; ZFILE ip
; struct bucket* buck = NULL
; if (!(ip = myfopen(fn, ZRMODE, 1)))
return 1
; const char* dochandle[6] = { "barcode", "3p-ad", "3p-si", "5p-si", "tabu", "filetag" }
; X_CASCADE_CLEAN(table_read(ip, thetable, 's', TABLE_WITH_COLNAMES), myfzclose(ip, 1))
; X_CASCADE(buckets_alloc(pam, thetable->n_rows))
; buck = pam->buck
; id_barcode = table_column_index(thetable, dochandle[doc_barcode])
; id_3p_ad = table_column_index(thetable, dochandle[doc_3p_ad])
; id_3p_si = table_column_index(thetable, dochandle[doc_3p_si])
; id_5p_si = table_column_index(thetable, dochandle[doc_5p_si])
; id_tabu = table_column_index(thetable, dochandle[doc_tabu])
; id_filetag = table_column_index(thetable, dochandle[doc_filetag])
; n_columns_found
= ( id_3p_ad >= 0 )
+ ( id_barcode >= 0 )
+ ( id_3p_si >= 0 )
+ ( id_5p_si >= 0 )
+ ( id_tabu >= 0 )
+ ( id_filetag >= 0 )
; switch(mode)
{ case CASE_NOBC
: { X_CASCADE(column_require(id_3p_ad, dochandle[doc_3p_ad]))
; X_CASCADE(column_require(id_tabu, dochandle[doc_tabu]))
; if (n_columns_found > 2)
argh("reaper", "only using columns %s and %s", dochandle[doc_3p_ad], dochandle[doc_tabu])
; break
; }
case CASE_3P_BC
: { X_CASCADE(column_require(id_3p_si, dochandle[doc_3p_si]))
; X_CASCADE(column_require(id_barcode, dochandle[doc_barcode]))
; X_CASCADE(column_require(id_tabu, dochandle[doc_tabu]))
; X_CASCADE(column_require(id_3p_ad, dochandle[doc_3p_ad]))
; if (n_columns_found > 4)
argh("reaper", "only using columns %s, %s and %s", dochandle[doc_barcode], dochandle[doc_3p_si], dochandle[doc_tabu])
; break
; }
case CASE_5P_BC
: { X_CASCADE(column_require(id_barcode, dochandle[doc_barcode]))
; X_CASCADE(column_require(id_5p_si, dochandle[doc_5p_si]))
; X_CASCADE(column_require(id_3p_ad, dochandle[doc_3p_ad]))
; X_CASCADE(column_require(id_tabu, dochandle[doc_tabu]))
; if (n_columns_found > 4)
argh
( "reaper"
, "only using columns %s, %s, %s and %s"
, dochandle[doc_barcode]
, dochandle[doc_5p_si]
, dochandle[id_3p_ad], dochandle[doc_tabu]
)
; break
; }
default
: ;
}
; for (n_barcodes=0; n_barcodes < thetable->n_rows; n_barcodes++)
{ int n = n_barcodes
; buck[n].barcode5p = NULL
; buck[n].sinsert5p = NULL
; buck[n].sinsert5p_n = 0
; buck[n].barcode3p = NULL
; buck[n].sinsert3p = NULL
; buck[n].adaptor3p = NULL
; buck[n].sinsert3p_n = 0
; buck[n].cat3p = NULL
; buck[n].filetag = NULL
; buck[n].barcode_n = 0
#define hte(s) (strcmp(s, "-") ? s : "") /* hyphen-to-empty */
; if (id_filetag >= 0)
buck[n].filetag = hte(table_elem_string(thetable, n, id_filetag))
; if (mode == CASE_NOBC)
{ buck[n].adaptor3p = hte(table_elem_string(thetable, n, id_3p_ad))
; parse_tabu(&buck[n].tabu2, hte(table_elem_string(thetable, n, id_tabu)), extent)
; if (id_5p_si >= 0)
buck[n].sinsert5p = hte(table_elem_string(thetable, n, id_5p_si))
, buck[n].sinsert5p_n = strlen(buck[n].sinsert5p) /* fixme do length assignments in unified place */
; }
else if (mode == CASE_3P_BC)
{ buck[n].barcode3p = table_elem_string(thetable, n, id_barcode)
; buck[n].adaptor3p = hte(table_elem_string(thetable, n, id_3p_ad))
; buck[n].sinsert3p = hte(table_elem_string(thetable, n, id_3p_si))
; parse_tabu(&buck[n].tabu2, hte(table_elem_string(thetable, n, id_tabu)), extent)
; buck[n].sinsert3p_n = strlen(buck[n].sinsert3p)
; buck[n].barcode_n = strlen(buck[n].barcode3p)
; buck[n].cat3p = stringle("%s%s%s", buck[n].sinsert3p, buck[n].barcode3p, buck[n].adaptor3p)
; if (buck[n].sinsert3p_n > 1)
X_FINISHED("3p sequence insert of size > 1 not yet supported; ask Stijn")
; if (n == 0)
barcodes_length = strlen(buck[n].barcode3p)
; else if (barcodes_length != strlen(buck[n].barcode3p))
argh("reaper", "barcodes with different lengths")
/* X_FINISHED("barcodes with different lengths") */
; }
else if (mode == CASE_5P_BC)
{ buck[n].barcode5p = table_elem_string(thetable, n, id_barcode)
; buck[n].sinsert5p = hte(table_elem_string(thetable, n, id_5p_si))
; buck[n].adaptor3p = hte(table_elem_string(thetable, n, id_3p_ad))
; parse_tabu(&buck[n].tabu2, hte(table_elem_string(thetable, n, id_tabu)), extent)
; buck[n].sinsert5p_n = strlen(buck[n].sinsert5p)
; buck[n].barcode_n = strlen(buck[n].barcode5p)
; if (n == 0)
barcodes_length = strlen(buck[n].barcode5p)
; else if (barcodes_length != strlen(buck[n].barcode5p))
argh("reaper", "barcodes with different lengths")
/* X_FINISHED("barcodes with different lengths") */
; }
if (extent > 0)
{ if (buck[n].adaptor3p && strlen(buck[n].adaptor3p) > extent)
((char*) buck[n].adaptor3p)[extent] = '\0'
; }
argh
( "reaper"
, "scanned line %d sinsert3p [%s] barcode3p [%s] adaptor3p [%s] cat3p [%s]"
" barcode5p [%s] sinsert5p [%s] tabu [%s]"
, (int) (n+1)
, buck[n].sinsert3p ? buck[n].sinsert3p : "NA"
, buck[n].barcode3p ? buck[n].barcode3p : "NA"
, buck[n].adaptor3p ? buck[n].adaptor3p : "NA"
, buck[n].cat3p ? buck[n].cat3p : "NA"
, buck[n].barcode5p ? buck[n].barcode5p : "NA"
, buck[n].sinsert5p ? buck[n].sinsert5p : "NA"
, buck[n].tabu2.tabuseq ? buck[n].tabu2.tabuseq : ""
)
; { struct atabu* boo = buck[n].tabu2.next
; while (boo)
{ fprintf(stderr, " [%s]", boo->tabuseq)
; boo = boo->next
; }
}
}
; return 0
; }
/* is the below comment up to date? */
/* The second condition is because we only fill part of the alignment matrix.
* in the equality case the matrix will be emtpy, and sw_trace3 will crash.
* The easiest way is to check here (BUT, fixme nevertheless; sw_trace3 should
* do better; it's probably crashing because max_ij is never increased and starts at 0).
* Note in barcode mode we need at least the barcode (plus a little bit more,
* likely), so an initial read of length cleanlength (before aligning)
* is never going to make it.
*/
int check_for_tabu
( struct record* rec
, unsigned len
, struct param* pam
, struct atabu* boo
)
{ struct sw_alninfo ai = { 0 }
; char* s = rec->seq
; char* q = rec->q
; int indel_allowed = pam->mr_tabu.mr_maxgap > 0
; char c_s = s[len] /* hack/kludge; note that len is user specified */
; char c_q = q[len] /* hack/kludge NOTE harmless in case q_n == 0 */
; int found_alignment = 0
; SWNUM data[MATRIX_SIZE] = { 0 }
; if
( !boo
|| !boo->tabulen
|| !pam->mr_tabu.mr_minlen
)
return 0
; s[len] = '\0'
; q[len] = '\0'
; do
{ if (sw_fill2(&ai, data, MATRIX_SIZE, s, boo->tabuseq, indel_allowed, &pam->swp_tabu))
X_IGNORE("sw error in check_for_tabu")
; rec->rs.N_cells += ai.n_cells_used
; sw_trace3(&ai, &pam->swp_tabu, ai.max_ij, SW_TRIM) /* check_for_tabu */
; if (sw_match_any(&ai, &pam->mr_tabu))
{ found_alignment = 1
; break
; }
boo = boo->next
; }
while (boo && boo->tabulen)
/* in check_for_tabu, fplint; this one printed before read */
; if
( pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
sw_pp2("read", x_tabu[found_alignment], -1, &ai, pam->fplint, pam->zippit, rec->rs.N, NULL)
/* in check_for_tabu */
; s[len] = c_s /* hack/kludge */
; q[len] = c_q /* hack/kludge */
; return found_alignment
; }
/* note: mr_minlen not used! */
/* note: sinsert is left in the alignment */
int sw_match_5p_required
( struct sw_alninfo* ai
, int adaptor_length
, struct match_requirement* mr
)
{ int n_edit = ai->n_insl + ai->n_insr + ai->n_subst
; int n_border_left = THEMAX(ai->lft_start, ai->rgt_start) - 1
; int n_border_right = adaptor_length - ai->lft_end
; int n_border = n_border_left + n_border_right
;if(0)fprintf(stderr, "border-left: %d border-right: %d\n", n_border_left, n_border_right)
; return
( ai->n_match + n_edit + n_border >= adaptor_length /* cover entire adapter */
&& n_edit + n_border <= mr->mr_maxedit /* edit requriement */
&& ai->n_insl + ai->n_insr <= mr->mr_maxgap /* gap requirement */
)
? 1
: 0
; }
/* only called by ntnt */
int check_5p_sinsert
( struct record* rec
, struct param* pam
, ZFILE fpclean /* clean file is barcode specific */
, const char** msgp
)
{ int keepall = pam->modes & MODE_KEEPALL
; if
( !pam->sinsert_5p
|| !strlen(pam->sinsert_5p)
)
return 0
;if(0)fprintf(stderr, "insert %s\n", pam->sinsert_5p)
; pam->swp_5p.right_limit = strlen(pam->sinsert_5p) + pam->mr_5p_require.mr_offset
; { struct sw_alninfo ai = { 0 }
; SWNUM data[MATRIX_SIZE] = { 0 }
; int adaptor_length = strlen(pam->sinsert_5p)
; int indel_allowed = pam->mr_5p_require.mr_maxgap > 0
; int found_alignment = 0
; if (sw_fill2(&ai, data, MATRIX_SIZE, pam->sinsert_5p, rec->seq, indel_allowed, &pam->swp_5p))
X_FINISHED_VAL(-1, "sw error in check_5p_sinsert")
; rec->rs.N_cells += ai.ni * ai.nj
; sw_trace3
( &ai
, &pam->swp_5p
, ai.max_ij
, indel_allowed ? SW_TRIM : 0
) /* check_5p_sinsert */
; found_alignment = sw_match_5p_required(&ai, adaptor_length, &pam->mr_5p_require)
;if(0)
fprintf(stderr, "%s %s %d\n", pam->sinsert_5p, rec->seq, found_alignment)
,sw_pp2("read", x_5psi[found_alignment], -1, &ai, stdout, 0, rec->rs.N, NULL)
/* in check_5p_sinsert, fpclean, fplint; caller ntnt returns immediately */
; if
( pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
sw_pp2("read", x_5psi[found_alignment], -1, &ai, pam->fplint, pam->zippit, rec->rs.N, NULL)
/* in check_5p_sinsert */
; if (found_alignment)
return ai.rgt_end + (ai.ni-1-ai.lft_end) /* ai.ni-1 is the length of left sequence */
; else
{ if (keepall)
{ msgp[0] = "5p-sinsert"
; return rec->seq_n
; }
rec->rs.N_5p_nosinsert++
; rec->rs.N_lint++
; rec->out_message = "5p-absent"
; dump_read(0, pam->fplint, rec, NULL, NULL, pam)
; return -1
; }
}
return 0
; }
/* fixme; what if the input read has length smaller than cleanlength and no
* adaptor or tabu is found? detected?
*/
int ntnt /* nucleotide nip and tuck */
( struct record* rec
, struct param* pam /* lint file is taken from pam, non-specific */
, ZFILE fpclean /* clean file is barcode specific */
)
{ int clean = 0, lint_offset = 1, p_5p_offset = 0
; int indel_allowed = pam->mr_3p_global.mr_maxgap > 0
; int keepall = pam->modes & MODE_KEEPALL
; const char* msg = "clean"
; int found_alignment = 0
; struct sw_alninfo ai = { 0 }
; struct match_account ma = { 'n', 0, 0, -1, -1 }
; struct sw_param myswp = pam->swp_3p
; SWNUM data[MATRIX_SIZE] = { 0 }
/* returns the offset of the first good bit */
/* updates rec->rs */
; p_5p_offset = check_5p_sinsert(rec, pam, fpclean, &msg)
; if (p_5p_offset < 0)
return 1
; else if (p_5p_offset > 0)
rec_remove_prefix(rec, p_5p_offset)
; if (g_guard > 0)
myswp.left_skip = g_guard
/* hierverder: if empty read, everything below is still acted
* upon; do we hit any of those conditions?
*/
/* CASE_NOBC may or may not set pam->adaptor_3p and pam->tabu2
* fixme check.
*/
; if (pam->adaptor_3p)
{ if
( sw_fill2
( &ai
, data
, MATRIX_SIZE
, rec->seq
, pam->adaptor_3p
, indel_allowed
, &myswp
) )
X_FINISHED("sw error in nip_and_tuck")
; rec->rs.N_cells += ai.ni * ai.nj
; sw_trace3(&ai, &myswp, ai.max_ij, SW_TRIMLEFT) /* decideme: why not SW_TRIMRIGHT? */
; }
ma.nnn_offset = get_nnn_offset(ai.left, ai.ni-1, pam->nnn_count, pam->nnn_outof)
; ma.qqq_offset = get_qqq_offset(rec, pam->q_cutoff, pam->q_winlen, pam->q_winofs, pam->q_readofs)
; ma.bbb_offset = pam->modes & MODE_QC_LATE ? get_bbb_offset(rec) : -1
; ma.aaa_offset = get_polya_offset(ai.left, ai.ni-1, pam->polya)
; ma.dust_offset= get_dust_offset(ai.left, ai.ni-1, pam->taildust, pam->taildust_bases)
/* NOTE everything in alignment space is offset one relative to string
* space (C's normal 0-based indexing). In alignment space the
* alignment starts at ai.lft_start, so the good bit ends at
* ai.lft_start-1. In string space that's at ai.lft_start-2. The
* length of the good bit is thus ai.lft_start - 1: that's what we
* need to compare with pam->cleanlength.
* NNN space is just string-based, so lint_offset is not adjusted when
* set to ma.nnn_offset. The same holds for BBB space.
* Finally, if 0 == pam->cleanlength, then we should always accept the
* read, even though the length cutoff comparison might *still* fail
* due to the ai.rgt_start offsetting -- we may have a clean
* readlength of -2 if for instance 3 == rgt_start.
*/
; if (pam->adaptor_3p)
found_alignment = sw_match_any(&ai, &pam->mr_3p_global)
; if
( pam->adaptor_3p[0] /* fixme/docme */
&& pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
sw_pp2("read", x_3pgb[found_alignment], -1, &ai, pam->fplint, pam->zippit, rec->rs.N, NULL)
; if (found_alignment)
{ ma.match_type = 'g'
; clean = (ai.lft_start + 1) > pam->cleanlength + ai.rgt_start
; msg = "len-global"
; if (clean)
lint_offset = (ai.lft_start + 1) - ai.rgt_start
; else if (!pam->cleanlength && ai.lft_start < ai.rgt_start)
clean = 1
, lint_offset = 1
; else
rec->rs.N_lenlo++
;
}
/* below, first use the overal best alignment.
* This may find a prefix match followed by low complexity sequence.
* Then look at the tail of the read only.
*/
else if (pam->adaptor_3p && pam->mr_3p_prefix.mr_minlen)
{ unsigned z1 = sw_match_end(&ai, &ma, &myswp, pam->p_3p_tth, &pam->mr_3p_prefix)
;if(0)fprintf(stderr, "z1 %d\n", (int) z1)
; found_alignment = z1 && ma.suffix_complexity < pam->suffix_complexity_max
; if (!found_alignment)
{ sw_trace3(&ai, &myswp, ai.max_ej, SW_TRIMLEFT) /* DON'T trim right; it needs to stay anchored */
; found_alignment = sw_match_end(&ai, &ma, &myswp, pam->p_3p_tth, &pam->mr_3p_prefix)
; if
( pam->adaptor_3p[0] /* fixme/docme */
&& pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
sw_pp2("read", x_3ppx[found_alignment], -1, &ai, pam->fplint, pam->zippit, rec->rs.N, NULL)
;if(0)fprintf(stderr, "found_alignment %d\n", (int) found_alignment)
; }
if (found_alignment)
{ if (ma.n_perfect)
clean = ai.ni > pam->cleanlength + ma.n_perfect
, ma.match_type = 't'
; else
clean = (ai.lft_start + 1) > pam->cleanlength + ai.rgt_start
, ma.match_type = 'p'
; msg = "len-suffix"
; if (clean)
{ lint_offset
= ma.n_perfect
? ai.ni - ma.n_perfect
: (ai.lft_start + 1) - ai.rgt_start
; }
else if (!pam->cleanlength && ai.lft_start < ai.rgt_start)
{ clean = 1
; lint_offset = 1
; }
else
rec->rs.N_lenlo++
; }
else if (ai.ni > pam->cleanlength)
clean = 1
, lint_offset = ai.ni
; }
else
clean = 1
, lint_offset = rec->seq_n + 1
/* This may happen due to the correction for right start
* (the alignment can start at base 2 in the adapter,
* base 0 in the read).
* Set to 1 by way of defensive programming.
*/
; if (!lint_offset)
lint_offset = 1
/* fixme: funcify */
; if (clean && ma.nnn_offset > 0)
{ rec->rs.B_NN += rec->seq_n+1 - ma.nnn_offset
; if (ma.nnn_offset <= pam->cleanlength)
{ clean = 0
; rec->rs.N_nn++
; msg = "NNN"
; }
else if (ma.nnn_offset < lint_offset)
lint_offset = ma.nnn_offset
; }
/* fixme: funcify */
if (clean && ma.qqq_offset > 0)
{ rec->rs.B_QQ += rec->seq_n+1 - ma.qqq_offset
; if (ma.qqq_offset <= pam->cleanlength)
{ clean = 0
; rec->rs.N_qq++
; msg = "lowmq"
; }
else if (ma.qqq_offset < lint_offset)
lint_offset = ma.qqq_offset
; }
/* ach, require 3 bases, why not */
if (pam->taildustlate && clean && lint_offset > 3)
{ int dust_offset = get_dust_offset(ai.left, lint_offset-1, pam->taildustlate, pam->taildustlate_bases)
; if (dust_offset >= 0)
{ if (ma.dust_offset < 0 || dust_offset < ma.dust_offset)
ma.dust_offset = dust_offset
; }
}
/* fixme: funcify */
if (clean && ma.dust_offset > 0)
{ rec->rs.B_tri += rec->seq_n+1 - ma.dust_offset
; if (ma.dust_offset <= pam->cleanlength)
{ clean = 0
; rec->rs.N_tri++
; msg = "tritail"
; }
else if (ma.dust_offset < lint_offset)
lint_offset = ma.dust_offset
; }
/* fixme: funcify */
if (clean && ma.bbb_offset > 0)
{ rec->rs.B_BB += rec->seq_n+1 - ma.bbb_offset
; if (ma.bbb_offset <= pam->cleanlength)
{ clean = 0
; rec->rs.N_bb++
; msg = "lowq"
; }
else if (ma.bbb_offset < lint_offset)
lint_offset = ma.bbb_offset
; }
/* fixme: funcify
* hierverder: is B_AA incremenent OK? (doublecounting)
*/
if (clean && ma.aaa_offset > 0)
{ rec->rs.B_AA += rec->seq_n+1 - ma.aaa_offset
; if (ma.aaa_offset <= pam->cleanlength)
{ clean = 0
; rec->rs.N_aa++
; msg = "polya"
; }
else if (ma.aaa_offset < lint_offset)
lint_offset = ma.aaa_offset
;if(0)fprintf(stderr, "polya %d lint %d clean %d seq %s %d\n", (int) ma.aaa_offset, (int) lint_offset, (int) clean, rec->seq, (int) rec->rs.N)
; }
/* fixme: funcify */
if (clean && pam->length_co > 0)
{ rec->rs.B_lco += rec->seq_n - pam->length_co
; if (pam->length_co+1 < lint_offset)
lint_offset = pam->length_co+1
; }
if (clean && pam->trintco && pam->trintco < trintscore(rec->seq, lint_offset-1))
{ clean = 0
; rec->rs.N_tri++
; msg = "trint"
; }
/* caller sets pam->the_tabu2 */
if
( clean
&& pam->the_tabu2.tabulen
&& check_for_tabu(rec, lint_offset-1, pam, &pam->the_tabu2)
)
{ if (keepall)
lint_offset = 1
; else
{ clean = 0
; rec->rs.N_tabu++
; }
msg = "tabu"
; }
clean |= keepall
; if (clean)
{ rec->rs.N_clean++
; rec->clean_n = lint_offset - 1
; rec->out_message = msg
; dump_read(1, fpclean, rec, &ai, &ma, pam)
; }
else
{ rec->rs.N_lint++
; if (!msg)
msg = "no message (weird)"
, rec->rs.N_nomsg++
; rec->out_message = msg
; dump_read(0, pam->fplint, rec, &ai, &ma, pam)
; rec->clean_n = -1
; }
return 0
; }
/* fixfixme: adaptor settings */
int do_adaptor_3p
( struct record* rec
, struct param* pam
)
{ pam->adaptor_3p = pam->buck[0].adaptor3p
; pam->the_tabu2 = pam->buck[0].tabu2
; pam->sinsert_5p = pam->buck[0].sinsert5p
#if 0
= (pam->buck[bestid].sinsert5p_n)
? pam->buck[bestid].sinsert5p
: NULL
#endif
; X_CASCADE(ntnt(rec, pam, pam->fpclean))
; if (rec->clean_n >= 0)
qc_tally_out(rec, pam, pam->buck_n, rec->clean_n) /* /specialbucket/ */
; return 0
; }
int do_barcode_3p
( struct record *rec
, struct param *pam
)
{ int NNN_offset = get_nnn_offset(rec->seq, rec->seq_n, pam->nnn_count, pam->nnn_outof)
; int BBB_offset = pam->modes & MODE_QC_LATE ? get_bbb_offset(rec) : -1
; int lint_offset = 0
; int indel_allowed = pam->mr_3p_global.mr_maxgap > 0
; SWNUM data[MATRIX_SIZE] = { 0 }
; struct sw_alninfo ai = { 0 }
; int buckid, id_clean = 0, n_clean_best = 0
; const char* msg = "nomatch"
; int score_best = 0, n_pass = 0
; const char* cfl[10] = { "cfl0", "cfl1", "cfl2", "cfl3", "cfl4", "cfl5", "cfl6", "cfl7", "cfl8", "cflZ" }
; for (buckid = 0; buckid < pam->buck_n; buckid++)
{ int score = 0
/* NOTE:
* We used to do restriction of search space
* (e.g. rec->seq + pam->cleanlength, limit
* search to start of adaptor). That restriction was dropped,
* but the reasons for dropping were not noted unfortunately.
*/
; if
( sw_fill2
( &ai
, data
, MATRIX_SIZE
, rec->seq
, pam->buck[buckid].cat3p
, indel_allowed
, &pam->swp_3p
) )
X_FINISHED("sw error in do_barcode_3p")
; rec->rs.N_cells += ai.ni * ai.nj
; sw_trace3(&ai, &pam->swp_3p, ai.max_ij, 0) /* do_barcode_3p */
; lint_offset = 1
; if ((ai.lft_start + 1) > ai.rgt_start)
lint_offset = (ai.lft_start + 1) - ai.rgt_start
/* donow: ship barcode offset and length */
; if
( sw_match_bc3p
( &ai
, &pam->mr_3p_global
, &pam->mr_3p_prefix
, &pam->mr_3p_barcode
, pam->buck[buckid].sinsert3p_n
, pam->buck[buckid].barcode_n
, &score /* ignored for now */
) )
{ score = ai.n_match - ai.n_insl - ai.n_insr - ai.n_subst
; n_pass++
; if (score > score_best)
{ id_clean = buckid
; n_clean_best = 1
; score_best = score
; }
else if (score == score_best)
n_clean_best++
; pam->buck[buckid].cache_score = score
; pam->buck[buckid].cache_accept = 1
; }
else
lint_offset = ai.ni
, pam->buck[buckid].cache_score = score
, pam->buck[buckid].cache_accept = 0
; pam->buck[buckid].cache_lint_offset = lint_offset
; }
if (n_clean_best == 0)
rec->rs.N_nomatch++
; if (n_clean_best == 1 && NNN_offset > 0)
{ rec->rs.B_NN += rec->seq_n+1 - NNN_offset
; if (NNN_offset <= pam->cleanlength)
{ n_clean_best = 0
; rec->rs.N_nn++
; msg = "NNN"
; }
else if (NNN_offset < lint_offset)
lint_offset = NNN_offset
; }
if (n_clean_best == 1 && BBB_offset > 0)
{ rec->rs.B_BB += rec->seq_n+1 - BBB_offset
; if (BBB_offset <= pam->cleanlength)
{ n_clean_best = 0
; rec->rs.N_bb++
; msg = "lowq"
; }
else if (BBB_offset < lint_offset)
lint_offset = BBB_offset
; }
/* The check for pam->cleanlength is so that we do
* not drop reads that have a virtual 'negative'
* offset of the first base of aligned adaptor.
*/
if
( n_clean_best == 1
&& pam->cleanlength
&& ai.lft_start + 1 <= pam->cleanlength + ai.rgt_start
)
{ n_clean_best = 0
; msg = "len"
; rec->rs.N_lenlo++
; }
/* Soft assertion: see comment below */
if (n_clean_best == 1 && !lint_offset)
{ argh("reaper", "nasty zero lint offset, please fix code")
; lint_offset = 1
; }
if
( n_clean_best == 1
&& check_for_tabu(rec, lint_offset-1, pam, &pam->buck[id_clean].tabu2)
/* warning: in the above check (and further below as well)
* we need that lint_offset > 0.
* this should be garantueed by having a match and
* by restricting the search space, but
* with an all-zero alignment matrix it could
* still be icky. So there's an assertion above.
*/
)
{ n_clean_best = 0
; rec->rs.N_tabu++
; msg = "tabu"
; }
if (n_clean_best == 1)
{ rec->clean_n = pam->buck[id_clean].cache_lint_offset - 1
; rec->out_message = NULL
; dump_read(1, pam->buck[id_clean].fp, rec, NULL, NULL, pam)
; rec->rs.N_clean++
#if 0
; if (n_pass > 1)
rec->rs.N_cflr++
#endif
; qc_tally_in(rec, pam, id_clean, rec->seq_n)
; qc_tally_out(rec, pam, id_clean, rec->clean_n)
; }
else
{ rec->out_message = n_clean_best ? cfl[n_clean_best > 9 ? 9 : n_clean_best] : msg
; dump_read(0, pam->fplint, rec, NULL, NULL, pam)
; qc_tally_out(rec, pam, pam->buck_n, rec->seq_n) /* specialbucket */
; rec->rs.N_lint++
; if (n_clean_best > 1)
rec->rs.N_cfl++
; }
/* in do_barcode_3p, buck[].fp, fplint */
if
( pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
{ for (buckid = 0; buckid < pam->buck_n; buckid++)
{ int score = 0
; if
( sw_fill2
( &ai
, data
, MATRIX_SIZE
, rec->seq
, pam->buck[buckid].cat3p
, indel_allowed
, &pam->swp_3p
) )
X_FINISHED("sw error in do_barcode_3p")
; rec->rs.N_cells += ai.ni * ai.nj
; sw_trace3(&ai, &pam->swp_3p, ai.max_ij, 0) /* do_barcode_3p */
; sw_match_bc3p
( &ai
, &pam->mr_3p_global
, &pam->mr_3p_prefix
, &pam->mr_3p_barcode
, pam->buck[buckid].sinsert3p_n
, pam->buck[buckid].barcode_n
, &score /* ignored */
)
; sw_pp2
( "read"
, x_3pbc[pam->buck[buckid].cache_accept]
, pam->buck[buckid].cache_score
, &ai
, pam->fplint
, pam->zippit
, rec->rs.N
, NULL
)
/* in do_barcode_3p */
; }
}
return 0
; }
int n_zip_mismatch
( const char* s1
, const char* s2
)
{ int n = 0
; const char* a, *b
; for (a=s1,b=s2; *a && *b; a++, b++)
n += *a != *b
; return n
; }
int do_barcode_5p
( struct record *rec
, struct param *pam
)
{ int buckid = 0
; int bestid = -1
; struct match_requirement mr = pam->mr_5p_barcode
; int indel_allowed = mr.mr_maxgap > 0
; int bestend = -1
; int conflict = 0
; SWNUM data[MATRIX_SIZE] = { 0 }
; struct sw_alninfo ai = { 0 }
; { for (buckid = 0; buckid < pam->buck_n; buckid++)
{ if (!strncmp(pam->buck[buckid].barcode5p, rec->seq, pam->buck[buckid].barcode_n))
{ bestend = pam->buck[buckid].barcode_n
; break
; }
}
if (bestend > 0)
bestid = buckid
; }
/* mr.mr_offset & 1 indicates a 'zip' alignment,
* toe to toe and head to head aligned with only substitutions allowed.
*/
if (bestid < 0 && mr.mr_maxedit > 0 && (mr.mr_offset & 1))
{ for (buckid = 0; buckid < pam->buck_n; buckid++)
{ int z = INT_MAX
; if (pam->buck[buckid].barcode_n <= rec->seq_n)
z = n_zip_mismatch(pam->buck[buckid].barcode5p, rec->seq)
#if WE_USE_ZLIB
;if(0)gzprintf(pam->fplint, "code %s seq %s zip %d\n", pam->buck[buckid].barcode5p, rec->seq, z)
#endif
; if (z <= mr.mr_maxedit)
{ if (bestid >= 0)
{ conflict = 1
#if WE_USE_ZLIB
; gzprintf
( pam->fplint
, "> 5p-conflict-zip %s %s [%s/%d] recno %d\n"
, pam->buck[bestid].barcode5p
, pam->buck[buckid].barcode5p
, rec->seq
, (int) rec->seq_n
, (int) rec->rs.N
)
#endif
; bestid = -1
; break
; }
else
bestid = buckid
; }
}
if (!conflict && bestid >= 0)
bestend = pam->buck[bestid].barcode_n
; }
if (!conflict && bestid < 0 && (mr.mr_offset & 2) && mr.mr_maxedit)
{ int n_edit = 0
; int n_gap = 0
; for (buckid = 0; buckid < pam->buck_n; buckid++)
{ int found_alignment = 0
; pam->swp_5p.right_limit = pam->buck[buckid].barcode_n + mr.mr_maxgap
; if (rec->seq_n < pam->buck[buckid].barcode_n)
continue
; if
( sw_fill2
( &ai
, data
, MATRIX_SIZE
, pam->buck[buckid].barcode5p
, rec->seq
, indel_allowed
, &pam->swp_5p /* governs alignment algorithm, settings are very stable */
) )
X_FINISHED("sw error in do_barcode_5p")
; rec->rs.N_cells += ai.ni * ai.nj
; sw_trace3(&ai, &pam->swp_5p, ai.max_ij, 0) /* do_barcode_5p */
; n_gap = ai.n_insl + ai.n_insr
; n_edit = ai.n_subst + n_gap
; found_alignment
= ( (ai.lft_end - ai.lft_start) + 1 + mr.mr_maxedit >= pam->buck[buckid].barcode_n + n_edit /* + (ai.rgt_start - 1) */
&& mr.mr_maxedit >= n_edit
&& mr.mr_maxgap >= n_gap
)
;if(0)fprintf(stderr, "%d\n", (int) ((ai.lft_end - ai.lft_start) + 1))
;if(0)fprintf(stderr, "found %d bc %s (%d %d)\n", (int) found_alignment, pam->buck[buckid].barcode5p, (int) n_edit, (int) (ai.rgt_start -1))
/* in do_barcode_5p, buck[].fp, fplint */
; if
( pam->verbose & VB_ALIGN
&& rec->rs.N % pam->sample_lint == 0
)
sw_pp2("read", x_5pbc[found_alignment], -1, &ai, pam->fplint, pam->zippit, rec->rs.N, NULL)
/* in do_barcode_5p */
; if (found_alignment)
{ if (bestid >= 0)
{
#if WE_USE_ZLIB
; gzprintf(pam->fplint, "> 5p-conflict-align %s %s %s\n", pam->buck[bestid].barcode5p, pam->buck[buckid].barcode5p, rec->seq)
#endif
; bestid = -1
; conflict = 1
; break
; }
else
bestid = buckid
, bestend = ai.rgt_end + (pam->buck[bestid].barcode_n - ai.lft_end)
; }
}
}
if (bestid < 0)
{ rec->rs.N_lint++
; rec->out_message = conflict ? "conflict" : "nobc"
; qc_tally_out(rec, pam, pam->buck_n, rec->seq_n) /* specialbucket */
; dump_read(0, pam->fplint, rec, NULL, NULL, pam)
; if (conflict)
rec->rs.N_cfl++
; else
rec->rs.N_nomatch++
; }
else
{ qc_tally_in(rec, pam, bestid, rec->seq_n) /* tally in per-barcode */
#if WE_USE_ZLIB
;if(0)gzprintf(pam->fplint, "# %s %s\n", pam->buck[bestid].barcode5p, rec->seq)
#endif
; rec_remove_prefix(rec, bestend) /* inclusive end in 1-based offset == length [12345] */
; pam->sinsert_5p
= (pam->buck[bestid].sinsert5p_n)
? pam->buck[bestid].sinsert5p
: NULL
; pam->adaptor_3p = pam->buck[bestid].adaptor3p
; pam->the_tabu2 = pam->buck[0].tabu2
; X_CASCADE(ntnt(rec, pam, pam->buck[bestid].fp))
; if (rec->clean_n >= 0)
qc_tally_out(rec, pam, bestid, rec->clean_n)
; }
return 0
; }
int parse_mr
( const char* form
, struct match_requirement* mr
, int maxnums
)
{ int n_consumed, ok = 1
; mr->mr_maxgap = 0
; do
{ if
( maxnums >= 4
&& 4 == sscanf(form, "%u/%u/%u/%d%n",
&mr->mr_minlen, &mr->mr_maxedit, &mr->mr_maxgap, &mr->mr_offset, &n_consumed)
)
break
; if
( maxnums >= 3
&& 3 == sscanf(form, "%u/%u/%u%n",
&mr->mr_minlen, &mr->mr_maxedit, &mr->mr_maxgap, &n_consumed)
)
break
; if
( maxnums >= 2
&& 2 == sscanf(form, "%u/%u%n",
&mr->mr_minlen, &mr->mr_maxedit, &n_consumed)
)
break
; ok = 0
; }
while (0)
; if (ok && mr->mr_maxedit < mr->mr_maxgap)
{ argh
( "error"
, "maximum gap count %d should not exceed maximum edit distance %d in %s"
, ji mr->mr_maxgap
, ji mr->mr_maxedit
, form
)
; mr->mr_maxedit = mr->mr_maxgap
; argh("fixit", "setting maximum edit distance to %d", ji mr->mr_maxedit)
; }
return ok && n_consumed == strlen(form)
; }
int parse_anchor
( const char* form
, int *d
, const char **anchor
)
{ int n_consumed = 0
; if (1 != sscanf(form, "%d/%n", d, &n_consumed))
return 0
; if (n_consumed == 0 || form[n_consumed-1] != '/')
return 0
; d[0] -= 1
; anchor[0] = form + n_consumed
; return 1
; }
int parse_pair
( const char* form
, unsigned *u1
, unsigned *u2
)
{ int n_consumed
; if (2 != sscanf(form, "%u/%u%n", u1, u2, &n_consumed))
return 0
; if (n_consumed != strlen(form))
return 0
; return 1
; }
int parse_dust_spec
( const char* form
, unsigned *cutoff
, unsigned *bases
)
{ int n_converted = 0
; if
( 2 != sscanf(form, "%u/%n", cutoff, &n_converted)
&& 1 != sscanf(form, "%u", cutoff)
)
return 0
; if (n_converted > 0)
{ const char* o = form + n_converted
; if (strchr(o, 'A')) bases[0] |= 1 << 0
; if (strchr(o, 'C')) bases[0] |= 1 << 1
; if (strchr(o, 'G')) bases[0] |= 1 << 2
; if (strchr(o, 'T')) bases[0] |= 1 << 3
; }
return 1
; }
static volatile sig_atomic_t g_abort_loop = 0;
void reaper_sig_catch
( int sig
)
{ if (sig == SIGINT)
g_abort_loop = 1
; }
int reaper_main
( int argc
, const char* argv[]
)
{ ZFILE input = NULL
; const char* g_fnin = "-"
; char* fnclean = NULL, *fnlint = NULL
; unsigned g_limit = 0
; const char* g_format_in = "@%I%A%n%R%n+%#%Q%n"
; const char* g_format_in2 = NULL
; int status = 19
; int called_from_R = 0
/* below initialises read_stats_global (rec.rs) */
; struct record rec = { { 0 }, { 0 }, { 0 }, { 0 }, { 0 }, 0 }
; struct param pam = { 0 }
#ifdef WINDOWS_BUILD
; set_argh("reaperlog.txt", 1)
#endif
; param_init(&pam)
; g_abort_loop = 0
; signal(SIGINT, reaper_sig_catch)
; themap_init()
;
/* enter macromagicalitaciousness */
arg_switch()
optarg("-i") g_fnin = thearg(); endarg()
optarg("-fastq") g_fnin = thearg(); endarg()
optarg("-clean-length") pam.cleanlength = atoi(thearg()); endarg()
optarg("-polya") pam.polya = atoi(thearg()); endarg()
optarg("-dust-suffix")
if (!parse_dust_spec(thearg(), &pam.taildust, &pam.taildust_bases))
X_ERR_JUMP(DONE, "-dust-suffix argument needs <cut-off> or <cutoff>/<ACGT> pattern");
endarg()
optarg("-dust-suffix-late")
if (!parse_dust_spec(thearg(), &pam.taildustlate, &pam.taildustlate_bases))
X_ERR_JUMP(DONE, "-dust-suffix-late argument needs <cut-off> or <cutoff>/<ACGT> pattern");
endarg()
optarg("-qqq-check")
if
( 4 != sscanf(thearg(), "%u/%u/%u/%u", &pam.q_cutoff, &pam.q_winlen, &pam.q_winofs, &pam.q_readofs)
&& 3 != sscanf(thearg(), "%u/%u/%u", &pam.q_cutoff, &pam.q_winlen, &pam.q_winofs)
&& 2 != sscanf(thearg(), "%u/%u", &pam.q_cutoff, &pam.q_winlen)
)
X_ERR_JUMP(DONE, "-medq argument needs <cutoff>/<winlen>[/<winoffset>] pattern");
if (pam.q_winofs > pam.q_winlen) pam.q_winofs = pam.q_winlen;
if (pam.q_winlen > 255) pam.q_winlen = 255;
if (pam.q_winlen % 2 == 0) pam.q_winlen++;
if (!pam.q_winofs) pam.q_winofs = (pam.q_winlen + 1) / 2;
endarg()
optarg("-anchor")
if (!parse_anchor(thearg(), &pam.anchor_offset, &pam.anchor_string))
X_ERR_JUMP(DONE, "-anchor argument does not follow <OFFSET>/<BASES> pattern");
fprintf(stderr, "filtering on substring %s at position %d\n", pam.anchor_string, pam.anchor_offset+1);
endarg()
optarg("-tri")
pam.trintco = atoi(thearg());
if (pam.trintco < 0 || pam.trintco > 100) {
argh("reaper", "-trint option should take argument in [0..100]");
pam.trintco = 0;
}
endarg()
optarg("-mr-tabu")
if (!parse_mr(thearg(), &pam.mr_tabu, 4))
X_ERR_JUMP(DONE, "-mr-tabu argument needs x/y or x/y/z or x/y/z/u pattern");
endarg()
optarg("-5p-barcode")
if (!parse_mr(thearg(), &pam.mr_5p_barcode, 4))
X_ERR_JUMP(DONE, "-5p-barcode argument needs x/y or x/y/z or x/y/z/u pattern");
endarg()
optarg("-5p-sinsert")
if (!parse_mr(thearg(), &pam.mr_5p_require, 4))
X_ERR_JUMP(DONE, "-5p-sinsert argument needs x/y or x/y/z or x/y/z/u pattern");
endarg()
optarg("-3p-global")
if (!parse_mr(thearg(), &pam.mr_3p_global, 4))
X_ERR_JUMP(DONE, "-3p-global argument needs x/y or x/y/z or x/y/z/u pattern");
endarg()
optarg("-3p-prefix")
if (!parse_mr(thearg(), &pam.mr_3p_prefix, 4))
X_ERR_JUMP(DONE, "-3p-prefix argument needs x/y or x/y/z or x/y/z/u pattern");
endarg()
optarg("-3p-barcode")
if (!parse_mr(thearg(), &pam.mr_3p_barcode, 3))
X_ERR_JUMP(DONE, "-3p-barcode argument needs x/y or x/y/z pattern");
endarg()
optarg("-3p-head-to-tail") pam.p_3p_tth = atoi(thearg()); endarg() /* suffix perfect match */
optarg("-swp")
if (3 != sscanf(thearg(), "%u/%u/%u",
&pam.swp_3p.gain_match, &pam.swp_3p.cost_subst, &pam.swp_3p.cost_gapleft)
)
X_ERR_JUMP(DONE, "-swp argument needs MATCH/SUBS/GAP pattern (gain/cost/cost)");
pam.swp_3p.cost_gapright = pam.swp_3p.cost_gapleft;
pam.swp_5p = pam.swp_3p;
pam.swp_tabu = pam.swp_3p;
pam.swp_raptor= pam.swp_3p;
endarg()
optarg("-loco")
pam.loco_check = atof(thearg());
endarg()
optarg("-geom")
if (!strcmp(thearg(), "3p-bc"))
g_mode = CASE_3P_BC
; else if (!strcmp(thearg(), "5p-bc"))
g_mode = CASE_5P_BC
; else if (!strcmp(thearg(), "no-bc"))
g_mode = CASE_NOBC
; else
X_ERR_JUMP(DONE, "unsupported geometry [%s]", thearg())
;
endarg()
optarg("-meta") pam.fnmeta = thearg(); endarg()
optarg("-record-format2") g_format_in2 = thearg(); endarg()
optarg("-record-format") g_format_in = thearg(); endarg()
uniarg("--fasta-in") g_format_in = ">%I%A%n%R%n"; endarg()
uniarg("--fastqx-out")
pam.format_clean = "@%I%trecno=%J%n%C%n+%n%Q%n";
pam.format_lint = "@%I%trecno=%J%tmsg=%M%n%R%n+%n%Y%n";
endarg()
uniarg("--fasta-out")
pam.format_clean = ">%I%n%C%n";
pam.format_lint = ">%I%tmsg=%M%n%R%n";
endarg()
uniarg("--fastax-out")
pam.format_clean = ">%I%trecno=%J%n%C%n";
pam.format_lint = ">%I%trecno=%J%tmsg=%M%n%R%n";
endarg()
optarg("-sc-max") pam.suffix_complexity_max = atof(thearg()); endarg()
optarg("-nnn-check")
if (!parse_pair(thearg(), &pam.nnn_count, &pam.nnn_outof))
X_ERR_JUMP(DONE, "-nnn-check argument does not follow x/y pattern");
endarg()
optarg("-sample")
if (!parse_pair(thearg(), &pam.sample_clean, &pam.sample_lint))
X_ERR_JUMP(DONE, "-sample argument does not follow x/y pattern");
endarg()
optarg("-debug")
if (strchr(thearg(), 'a')) pam.verbose |= VB_ALIGN;
if (strchr(thearg(), 'c')) pam.verbose |= VB_CLEAN;
if (strchr(thearg(), 'l')) pam.verbose |= VB_LINT;
endarg()
optarg("-do") g_limit = atoi(thearg()); endarg()
uniarg("--nozip") pam.zippit = 0; endarg()
uniarg("--R") called_from_R = 1; endarg()
uniarg("--noqc") BIT_OFF(pam.modes, MODE_QC_FULL);endarg()
uniarg("--bcq-early") BIT_ON(pam.modes, MODE_QC_EARLY);endarg()
uniarg("--bcq-late") BIT_ON(pam.modes, MODE_QC_LATE); endarg()
uniarg("--full-length") BIT_ON(pam.modes, MODE_FULLLENGTH ); endarg()
uniarg("--keep-all") BIT_ON(pam.modes, MODE_KEEPALL ); endarg()
optarg("-3pa") pam.adaptor_3p = hte(thearg()); endarg()
optarg("-raptor") pam.adaptor_test = thearg(); endarg()
optarg("-kmer") pam.kmer = atoi(thearg()); endarg()
optarg("-5psi") pam.sinsert_5p = hte(thearg()); endarg()
optarg("-tabu") parse_tabu(&pam.the_tabu2, hte(thearg()), g_restrict); endarg()
optarg("-basename") pam.basename = thearg(); endarg()
optarg("-format-clean") pam.format_clean = thearg(); endarg()
optarg("-length-limit") pam.length_limit = atoi(thearg()); endarg()
optarg("-format-lint") pam.format_lint = thearg(); endarg()
optarg("-trim-length") pam.length_co = atoi(thearg()); endarg()
uniarg("--stream-clean") BIT_ON(pam.modes, MODE_STREAMCLEAN); endarg()
uniarg("--record-format")
fprintf(stdout, "-record-format <format string> (extended record description syntax)\n");
puts(" %R expect read (longest sequence found over [a-zA-Z]*) - empty read allowed");
puts(" %I expect identifier (longest sequence of non-blank)");
puts(" %A expect overflow annotation field (initial blanks skipped, rest of line)");
puts(" %a use previous %F or %G or %H field as annotation");
puts(" %B expect annotation field (longest sequence of non-blank)");
puts(" %Q expect quality (longest sequence of non-blank)");
puts(" %X expect count (a nonnegative integer number)");
puts(" %F expect and discard field (longest sequence of non-tab)");
puts(" %G expect and discard field (longest sequence of non-blank)");
puts(" %Hx (x is a placeholder) expect and discard up to x");
puts(" %# discard everything until end of line");
puts(" %b expect run of blanks (space or tab)");
puts(" %n expect end of line match");
puts(" %. expect and discard any character");
puts(" %s expect a space");
puts(" %t expect a tab");
puts(" %% expect a percent sign");
puts("Anything else requires a literal match, see the FASTA example just below");
puts("The string '>%I%#%R%n' will parse FASTA-ish input (but use --fasta-in instead)");
puts("The string '%F%t%R%#' retrieves the second field from each line");
status = 0;
goto DONE;
endarg()
optarg("-restrict")
g_restrict = atoi(thearg());
endarg()
optarg("-guard")
g_guard = atoi(thearg());
endarg()
uniarg("-z")
fprintf(stdout, "Struct param: %d\n", (int) sizeof(struct param));
fprintf(stdout, "Struct bucket: %d\n", (int) sizeof(struct bucket));
status = 0;
goto DONE;
endarg()
uniarg("--version")
fprintf(stdout, "Reaper version: %s\n", reaper_tag);
status = 0;
goto DONE;
endarg()
uniarg("-h")
puts("\nRequired options");
puts("-geom <mode> mode in {no-bc, 3p-bc, 5p-bc}");
puts("-meta <fname> file with geometry-dependent format. Required columns:");
puts(" Geometry Columns:");
puts(" no-bc 3p-ad - - - tabu");
puts(" 3p-bc 3p-ad barcode 3p-si - tabu");
puts(" 5p-bc 3p-ad barcode - 5p-si tabu");
puts("bc=barcode, ad=adaptor, si=sequence insert");
puts("Columns 3p-si, 5p-si, 3p-ad and tabu may all be empty");
puts("Alternatively, to express absence, a single hyphen may be used");
puts("\nImportant options");
puts("-i <fname> input stream (gzipped file allowed) (default STDIN)");
fprintf(stdout, "-clean-length <int> minimum allowed clean length (default %d)\n", (int) pam.cleanlength);
puts("-guard <int> protect first <int> bases in read from adapter and tabu matching");
fprintf(stdout, "-restrict <int> only use the first <int> bases of adapter or tabu sequence (default %d)\n", (int) g_restrict);
puts(" This is to avoid false positive matches");
puts("-tri <threshold> filter out reads with tri-nt score > threshold");
puts(" a reasonable <threshold> is 35");
fprintf(stdout, "-qqq-check <val>/<winlen> cut sequence when median quality drops below <val>\n");
fprintf(stdout, "-qqq-check <val>/<winlen>/<winofs> as above, cut at <winofs> (default %d)\n", (int) pam.q_winofs);
fprintf(stdout, "-qqq-check <val>/<winlen>/<winofs>/<readofs> as above, start at <readofs>\n");
fprintf(stdout, "-dust-suffix <threshold> dust theshold for read suffix (default %d, suggested 20)\n", (int) pam.taildust);
fprintf(stdout, "-nnn-check <count>/<outof> (default %d/%d)\n", (int) pam.nnn_count, (int) pam.nnn_outof);
puts(" disregard read onwards from seeing <count> N's in <outof> bases");
puts("\nAlignment options");
puts("Options to specify when part of an alignment triggers a match:");
#define local_EXPAND_MR(mr) (int) mr.mr_minlen, (int) mr.mr_maxedit, (int) mr.mr_maxgap, (int) mr.mr_offset
fprintf(stdout, "-3p-global l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_3p_global));
fprintf(stdout, "-3p-prefix l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_3p_prefix));
fprintf(stdout, "-3p-barcode l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_3p_barcode));
fprintf(stdout, "-5p-barcode l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_5p_barcode));
fprintf(stdout, "-5p-sinsert l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_5p_require));
fprintf(stdout, "-mr-tabu l/e[/g[/o]] (default %d/%d/%d/%d)\n", local_EXPAND_MR(pam.mr_tabu));
fprintf(stdout, "-3p-head-to-tail l minimal trailing perfect match length (default %d)\n", (int) pam.p_3p_tth);
#undef local_EXPAND_MR
puts(" syntax used in the above:");
puts(" l <int> minimum length required to count sub-alignment as match");
puts(" e <int> maximum allowed edit distance");
puts(" g <int> [optional, not active when set to 0] maximum allowed number of gaps");
puts(" o <int> [optional, not active when set to 0] offset:");
puts(" o= 5 requires alignment to start in the first five bases of adaptor");
puts(" o=-5 requires alignment to end in the last five bases of adaptor");
fprintf(stdout, "-swp M/S/G match/substitution/gap gain/cost/cost (default %u/%u/%u)\n",
pam.swp_3p.gain_match, pam.swp_3p.cost_subst, pam.swp_3p.cost_gapleft);
puts("\nInput/output options");
puts("--fasta-in read FASTA input");
fprintf(stdout, "-record-format <format string> (record description, default %s)\n", g_format_in);
puts(" [ -record-format syntax is output when supplying --record-format ]");
puts("-record-format2 <format string> (simple line formats, one field per line):");
puts(" R read");
puts(" I read identifier");
puts(" Q quality scores");
puts(" D discard field");
puts("");
puts("-basename <pfx> pfx.lint.gz, pfx.clean.gz pfx.report etc will be constructed");
puts("-format-clean <format string> (output for clean reads)");
puts("-format-lint <format string> (output for filtered reads)");
puts(" -format-clean/lint specification syntax:");
puts(" %R read");
puts(" %C clean read");
puts(" %Z clean read padded with Ns if necessary");
puts(" %V reverse complement of clean read");
puts(" %I read identifier");
puts(" %Q clean or input read quality (for clean / lint file respectively)");
puts(" %X read count (only applicable if -record-format is used)");
puts(" %Y input read quality");
puts(" %q<c> clean input read quality padded with character <c>");
puts(" %A annotation field");
puts(" %L clean read length");
puts(" %M message describing cause for filtering (lint file)");
puts(" %T trinucleotide complexity score (clean/lint file)");
puts(" %U dUst sUffix complexity information");
puts(" %3 best read/3p-adaptor alignment");
puts(" %= alignment characteristics");
puts(" mt=matchtype");
puts(" sc=suffix-complexity");
puts(" ht=head-tail-match");
puts(" nn=N-match-offset");
puts(" bb=B-match-offset");
puts(" aa=Polya-offset");
puts(" qq=Quality-trim-offset");
puts(" %n newline");
puts(" %J record offset, unique for each read. Use to match paired-end reads");
puts(" %f fastq line number based on standard fastq format");
puts(" %t tab");
puts(" %% percent sign");
puts(" Anything else is copied verbatim");
puts("-debug [acl]+ a=alignments c=clean l=lint");
puts("-sample c/l if debug, sample every c/l clean/lint read");
puts("--nozip do not output gzipped files");
puts("--noqc do not output quality report files");
puts("\nMiscellaneous options");
puts("--bcq-early perform early 'B' quality filtering (when reading sequences)");
puts("--bcq-late perform late 'B' quality filtering (before outputting sequences)");
puts("--full-length only allow reads not shortened in any filter step");
puts("--keep-all delete rather than discard reads (e.g. tabu match, missing 5p-sinsert)");
puts("-trim-length <int> cut reads back to length <int>");
fprintf(stdout, "-polya <int> remove trailing A's if length exceeds <int>\n");
fprintf(stdout, "-sc-max <f> threshold for complexity of suffix following prefix match (default %.2f)\n", pam.suffix_complexity_max);
puts("\nOptions for use when running reaper with -geom no-bc");
puts("-3pa <three prime adaptor>");
puts("-5psi <five prime sequence insert>");
puts("-tabu <tabu sequence>");
puts("");
status = 0;
X_ERR_JUMP(DONE, "DON'T PANIC");
endarg()
failarg()
arg_done()
/* curtains macromagicalitaciousness */
; { unsigned dummy
; kraken_readline(NULL, NULL, 0, NULL, &dummy) /* resets static buffers */
; }
if (called_from_R)
argh("reaper", "R is calling")
; if (g_mode == CASE_NOTSPECIFIED)
X_ERR_JUMP(DONE, "-geom option required (see -h for available options)")
; else if (g_mode > CASE_NOBC && !pam.fnmeta)
X_ERR_JUMP(DONE, "-meta option required with geometry %s", G_MODE[g_mode])
/* fixme: the logic of barcodes_io_open
* versus fpclean is somewhat entangled
* and dispersed throughout.
*/
; if (pam.fnmeta)
{ if (config_read(pam.fnmeta, &pam, g_mode, g_restrict))
X_ERR_JUMP(DONE, "failed to parse barcode spec file")
; if
( g_mode != CASE_NOBC
&& barcodes_io_open(pam.buck, pam.basename, pam.buck_n, pam.zippit)
)
X_ERR_JUMP(DONE, "error opening barcode bucket files")
; X_TRY_JUMP(DONE, buckets_prepare_qc(pam.buck, pam.buck_n, g_mode))
; if (g_mode == CASE_3P_BC)
{ pam.swp_3p.left_skip = pam.cleanlength
; pam.swp_3p.right_limit = pam.mr_3p_global.mr_minlen /* fixme document */
; pam.swp_3p.flags |= SW_ENDTOSTART
; pam.p_3p_tth = 0
; }
/* retireme/fixme: the below is spurious given call above. */
#if 0
/* fixme: this is a bit of a kludge: separate tally/qc stuff */
; if (g_mode == CASE_NOBC)
X_TRY_JUMP(DONE, buckets_prepare_qc(pam.buck, pam.buck_n, g_mode))
#endif
; }
else if (g_mode == CASE_NOBC)
{ X_TRY_JUMP(DONE, buckets_alloc(&pam, 1))
; X_TRY_JUMP(DONE, adaptors_set_case_3p(pam.buck, &pam.buck_n, &pam))
; X_TRY_JUMP(DONE, buckets_prepare_qc(pam.buck, pam.buck_n, g_mode))
/* note: in this branch we do not open io for (the one) bucket */
; }
else
X_ERR_JUMP(DONE, "-meta option is required for all modes except '3p-ad'")
; fnclean
= stringle
( "%s.%s.%s"
, pam.basename
, (g_mode == CASE_NOBC) ? "lane" : "ocean" /* ocean case should not materialise */
, pam.zippit ? "clean.gz" : "clean"
)
; fnlint = stringle("%s.%s", pam.basename, pam.zippit ? "lint.gz" : "lint")
;
if (!(input = myfopen(g_fnin, "r", 1)))
X_ERR_JUMP(DONE, "bailing out")
; if (g_mode == CASE_NOBC)
{ if (pam.modes & MODE_STREAMCLEAN)
fnclean = stringle("-")
; if (!(pam.fpclean = myfopen(fnclean, ZWMODE, pam.zippit)))
X_ERR_JUMP(DONE, "bailing out")
; }
if (!(pam.fplint = myfopen(fnlint, ZWMODE, pam.zippit)))
X_ERR_JUMP(DONE, "bailing out")
; { clock_t t1 = clock(), t2
; struct read_stats_global rsprev = { 0 } /* read stats per million */
; int anchor_len = 0
; if (pam.anchor_string)
anchor_len = strlen(pam.anchor_string)
; fprintf(stderr, "---\n")
; fprintf(stderr, "mRpm million reads per minute\n")
; fprintf(stderr, "mNpm million nucleotides per minute\n")
; fprintf(stderr, "mCps million alignment cells per second\n")
; fprintf(stderr, "lint total removed reads (per 10K), sum of columns to the left\n")
; fprintf
( stderr
, "%-37s %7s %3s %4s %4s %4s %6s%3s %4s %4s %4s %4s %4s %4s %4s %4s %4s%s\n"
, "25K reads per dot, 1M reads per line"
, "seconds", "mr", "mRpm", "mNpm", "mCps"
, "{error", "qc", "low", "len", "NNN", "tabu", "nobc", "cflr", "cfl", "lint", "OK", "} per 10K"
)
; do
{ unsigned rrstat = g_format_in2
? read_record2(input, g_format_in2, &rec, pam.length_limit)
: read_recordx(input, g_format_in, &rec)
; if (rec.seq_n > MATRIX_SIZE / 24)
rec.seq_n = MATRIX_SIZE / 24
, rec.seq[rec.seq_n] = '\0'
; if (rrstat == RR_DONE)
break
; else if (rrstat == RR_NOMEM || rrstat == RR_ERROR)
goto DONE
; if (pam.modes & MODE_FULLLENGTH)
pam.cleanlength = rec.seq_n
; rec.rs.B += rec.seq_n
; do
{ qc_tally_in(&rec, &pam, pam.buck_n, rec.seq_n) /* specialbucket */
; if (pam.anchor_offset >= 0 && pam.anchor_string)
{ if
( pam.anchor_offset <= rec.seq_n
&& strncmp(rec.seq+pam.anchor_offset, pam.anchor_string, anchor_len)
)
{ rec.rs.N_anchor++
; rec.rs.N_lint++
; break
; }
}
if (pam.modes & MODE_QC_EARLY)
{ int bbb_ofs
; if ((bbb_ofs = get_bbb_offset(&rec)) > 0)
{ rec.rs.B_BB += rec.seq_n + 1 - bbb_ofs
; if (bbb_ofs <= pam.cleanlength)
{ rec.rs.N_bb++; rec.rs.N_lint++; break ; }
else
{ rec.seq_n = bbb_ofs-1
; rec.q_n = bbb_ofs-1
; rec.seq[bbb_ofs-1] = '\0'
; rec.q[bbb_ofs-1] = '\0'
; }
}
}
if (pam.loco_check && loco(&rec, &pam, pam.loco_check))
{ rec.rs.N_loco++; rec.rs.N_lint++; break ; }
switch(g_mode)
{ case CASE_NOBC
: do_adaptor_3p(&rec, &pam); break
;
case CASE_3P_BC
: do_barcode_3p(&rec, &pam); break
;
case CASE_5P_BC
: do_barcode_5p(&rec, &pam); break
;
case CASE_NOTSPECIFIED
: X_ERR_JUMP(DONE, "impossible")
; }
/* pam->verbose & VB_ALIGN */
; }
while (0)
; if (25000 * (rec.rs.N / 25000) == rec.rs.N)
{ fputc('.', stderr)
; if (rec.rs.N % 1000000 == 0)
{ double nsecs
; t2 = clock()
; unsigned diffn = (rec.rs.N - rsprev.N)
; unsigned diffnt = (rec.rs.B - rsprev.B)
; unsigned diffsw = (rec.rs.N_cells - rsprev.N_cells)
; nsecs = (t2 - t1) * 1.0 / CLOCKS_PER_SEC
; fprintf
( stderr
, " %4.0f %3d %4.1f %4.0f %4.0f"
, nsecs
, (int) (rec.rs.N / 1000000)
, (double) (60.0 / nsecs)
, (double) (diffnt * 1.0 / (nsecs * 1000000.0 / 60.0))
, (double) (diffsw * 1.0 / (nsecs * 1000000.0))
)
; fprintf
( stderr
, " %4d %4d %4d %4d %4d %4d %4d %4d %4d %4d %4d\n"
, (int) (10000.5 * (rec.rs.N_err - rsprev.N_err) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_bb - rsprev.N_bb ) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_loco - rsprev.N_loco) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_lenlo - rsprev.N_lenlo) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_nn - rsprev.N_nn) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_tabu - rsprev.N_tabu) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_nomatch - rsprev.N_nomatch) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_cflr - rsprev.N_cflr) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_cfl - rsprev.N_cfl) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_lint - rsprev.N_lint) * 1.0 / diffn)
, (int) (10000.5 * (rec.rs.N_clean - rsprev.N_clean) * 1.0 / diffn)
)
; t1 = t2
; rsprev = rec.rs
; }
}
}
while (!g_abort_loop && (!g_limit || rec.rs.N < g_limit))
; fputc('\n', stderr)
; if
( rec.rs.N_bb + rec.rs.N_loco
+ rec.rs.N_lenlo + rec.rs.N_nn
+ rec.rs.N_tabu + rec.rs.N_cfl
+ rec.rs.N_5p_nosinsert + rec.rs.N_nomatch
+ rec.rs.N_aa
+ rec.rs.N_tri + rec.rs.N_qq
+ rec.rs.N_anchor
!= rec.rs.N_lint
)
argh("reaper", "(not overly important) discarded category counts do not add up to lint total")
; if (rec.rs.N_clean + rec.rs.N_lint + rec.rs.N_err != rec.rs.N)
argh("reaper", "(not overly important) dispatch category counts do not add up to input total")
; if (rec.rs.N_nomsg)
argh("reaper", "(not overly important) there were %u discarded reads with no message", rec.rs.N_nomsg)
; qc_brief(&pam, &rec, g_mode)
; if (g_fieldtosmall)
argh("reaper", "___ WARNING ___ %lu cases where field was truncated", (long unsigned) g_fieldtosmall)
; if (pam.modes & MODE_QC_FULL)
qc_report(&pam, &rec, g_mode)
; }
status = 0
; DONE:
; myfzclose(input, 1)
; myfzclose(pam.fpclean, pam.zippit)
; myfzclose(pam.fplint, pam.zippit)
; barcodes_io_close(pam.buck, pam.buck_n, pam.zippit)
; if (0) data_release(&pam)
; if (fnclean) free(fnclean)
; if (fnlint) free(fnlint)
; close_argh()
; return status
; }
#ifndef BUILD_R_BINDINGS
int main
( int argc
, const char* argv[]
)
{ return reaper_main(argc, argv)
; }
#endif
#if 0
#ifdef RREAPERR
SEXP R_dispatchee
( SEXP list
, int (*themain)(int , const char* [])
, const char* callee
)
{ /* SEXP elmt = R_NilValue */
; SEXP result
; SEXP names = getAttrib(list, R_NamesSymbol)
; int argc = 1
; const char* argv[100] = { 0 }
; argv[0] = "dummy-internal"
; int *res = NULL
; int i = 0
; PROTECT(result = allocVector(INTSXP, 1))
; res = INTEGER(result)
; if (length(list) <= 49) /* space for argv[0] and final NULL */
for (i = 0; i < length(list); i++)
{ const char* key = CHAR(STRING_ELT(names, i))
; const char* val = CHAR(STRING_ELT(list, i))
; if (!strlen(key))
{ if (!strncmp(val, "--", 2))
argv[argc++] = val
; else
{ argh(callee, "argument %s is not of type \"--mode\"", val)
; break
; }
}
else
{ if (!strncmp(key, "--", 2))
{ if (strlen(val) && strcmp(val, "on") && strcmp(val, "ON"))
{ argh(callee, "option %s=%s is not recognised as \"--mode=on\"", key, val)
; break
; }
argv[argc++] = key
; }
else
{ argv[argc++] = key
; argv[argc++] = val
; }
}
}
; argv[argc] = NULL
; if (i != length(list))
*res = 29
; else
{ int j
; fprintf(stderr, "Passing to %s:", callee)
; for (j=0;j<argc;j++)
fprintf(stderr, " %s", argv[j])
; fputc('\n', stderr)
; *res=themain(argc, argv)
; }
fputc('\n', stderr)
; UNPROTECT(1)
; return result
; }
SEXP reaperC
( SEXP list
)
{ return R_dispatchee(list, reaper_main, "reaper")
; }
#endif
#endif
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment