//this is designed to be used when loading multiple binaries into a
//single colour, which may have different metadata. In this case,
// we cumulate covg, get mean read length, and for each type of error cleaning we
//say yesif ANY of them did, taking the last threshold for any of them.
void graph_info_set_all_metadata(GraphInfo* target, GraphInfo* src, int colour, boolean dont_set_pool_cleaning)
{
graph_info_update_mean_readlen(target, colour,
target->mean_read_length[colour],
target->total_sequence[colour],
src->mean_read_length[colour],
src->total_sequence[colour]);
graph_info_increment_seq(target, colour, src->total_sequence[colour]);
target->seq_err[colour]=src->seq_err[colour];
error_cleaning_assign_with_OR(target->cleaning[colour], src->cleaning[colour], dont_set_pool_cleaning);
if ( (strcmp(src->sample_ids[colour], "undefined")!=0) //source has nontrivial sample id
&&
(strcmp(src->sample_ids[colour], target->sample_ids[colour])!=0) //src and target have different sample id
&&
(strcmp(target->sample_ids[colour],"undefined")!=0)//target has non trivial sample id
)
{
set_string_to_null(target->sample_ids[colour], MAX_LEN_SAMPLE_NAME );
strcat(target->sample_ids[colour], "pool");
}
else if (strcmp(src->sample_ids[colour], "undefined")!=0) //src has nontrivial sample id
{
//either it is the same, or target is undefined. Either way, this is ok:
set_string_to_null(target->sample_ids[colour], MAX_LEN_SAMPLE_NAME);
strcat(target->sample_ids[colour], src->sample_ids[colour]);
}
}
void graph_info_initialise_one_colour_except_pool_cleaning(GraphInfo* ginfo, int colour)
{
set_string_to_null(ginfo->sample_ids[colour], MAX_LEN_SAMPLE_NAME);
strcat(ginfo->sample_ids[colour], "undefined");
ginfo->sample_id_lens[colour]=strlen(ginfo->sample_ids[colour]);
graph_info_set_seq(ginfo, colour, 0);
graph_info_set_mean_readlen(ginfo, colour, 0);
ginfo->seq_err[colour]=0.01;
error_cleaning_initialise_except_pool_cleaning(ginfo->cleaning[colour]);
}
void error_cleaning_initialise(ErrorCleaning* cl)
{
error_cleaning_initialise_except_pool_cleaning(cl);
cl->cleaned_against_another_graph=false;
set_string_to_null(cl->name_of_graph_against_which_was_cleaned, MAX_FILENAME_LENGTH);
strcat(cl->name_of_graph_against_which_was_cleaned, "undefined");
cl->len_name_of_graph_against_which_was_cleaned=
strlen(cl->name_of_graph_against_which_was_cleaned);
}
void graph_info_initialise(GraphInfo* ginfo)
{
int i;
for (i=0; i<NUMBER_OF_COLOURS; i++)
{
set_string_to_null(ginfo->sample_ids[i], MAX_LEN_SAMPLE_NAME);
strcat(ginfo->sample_ids[i], "undefined");
ginfo->sample_id_lens[i]=strlen(ginfo->sample_ids[i]);
graph_info_set_seq(ginfo, i, 0);
graph_info_set_mean_readlen(ginfo, i, 0);
ginfo->seq_err[i]=0.01;
error_cleaning_initialise(ginfo->cleaning[i]);
}
}
//outputs fastQ unless add_greedy_bases_for_better_bwt_compression==true, in which case is for 1000genomes, and they want fastA
inline void error_correct_file_against_graph(char* fastq_file, char quality_cutoff, char ascii_qual_offset,
dBGraphEc *db_graph, char* outfile,
uint64_t *bases_modified_count_array,//distribution across reads; how many of the read_length bases are fixed
uint64_t *posn_modified_count_array,//where in the read are we making corrections?
int bases_modified_count_array_size,
int min_read_len,
HandleLowQualUncorrectable policy,
boolean add_greedy_bases_for_better_bwt_compression,
int num_greedy_bases,
boolean rev_comp_read_if_on_reverse_strand)
{
int max_read_len = bases_modified_count_array_size - 2;
int read_len_upper_bound = max_read_len + num_greedy_bases;
int read_len_lower_bound = min_read_len + num_greedy_bases;
int read_len_final = 0;
//reset the stats arrays, we get stats per input file
set_uint64_t_array(bases_modified_count_array,bases_modified_count_array_size, (uint64_t) 0);
set_uint64_t_array(posn_modified_count_array, bases_modified_count_array_size, (uint64_t) 0);
//set some variables, quality etc
quality_cutoff+=ascii_qual_offset;
short kmer_size = db_graph->kmer_size;
StrBuf* uncorrectedGoodQual_file = strbuf_create(outfile);
StrBuf* uncorrectedLowQual_file = strbuf_create(outfile);
StrBuf* corrected_file = strbuf_create(outfile);
StrBuf* discarded_undefined_file = strbuf_create(outfile);
StrBuf* discarded_uncorrectable_file = strbuf_create(outfile);
StrBuf* discarded_shortread_file = strbuf_create(outfile);
strbuf_append_str(uncorrectedGoodQual_file, ".printuncorrectedgoodqual");
strbuf_append_str(uncorrectedLowQual_file, ".printuncorrectedlowqual");
strbuf_append_str(corrected_file, ".printcorrected");
strbuf_append_str(discarded_undefined_file, ".discardundefinedbase");
strbuf_append_str(discarded_uncorrectable_file, ".discarduncorrectable");
strbuf_append_str(discarded_shortread_file, ".discardshortread");
FILE* uncorrectedGoodQual_fp = fopen(uncorrectedGoodQual_file->buff, "w");
FILE* uncorrectedLowQual_fp = fopen(uncorrectedLowQual_file->buff, "w");
FILE* corrected_fp = fopen(corrected_file->buff, "w");
FILE* discarded_undefined_fp = fopen(discarded_undefined_file->buff, "w");
FILE* discarded_uncorrectable_fp = fopen(discarded_uncorrectable_file->buff, "w");
FILE* discarded_shortread_fp = fopen(discarded_shortread_file->buff, "w");
char* suff1 = ".distrib_num_modified_bases";
char* suff2 = ".distrib_posn_modified_bases";
char* suff3 =".read_stats";
char* stat1 = (char*) malloc(sizeof(char)*(strlen(outfile)+strlen(suff1)+1));
char* stat2 = (char*) malloc(sizeof(char)*(strlen(outfile)+strlen(suff2)+1));
char* stat3 = (char*) malloc(sizeof(char)*(strlen(outfile)+strlen(suff3)+1));
if ( (stat1==NULL) || (stat2==NULL) || (stat3==NULL) )
{
die("Unable to malloc FILENAME strings. Something badly wrong with your server\n");
}
set_string_to_null(stat1, strlen(outfile)+strlen(suff1)+1);
set_string_to_null(stat2, strlen(outfile)+strlen(suff2)+1);
set_string_to_null(stat3, strlen(outfile)+strlen(suff3)+1);
strcpy(stat1, outfile);
strcat(stat1, suff1);
strcat(stat2, outfile);
strcat(stat2, suff2);
strcat(stat3, outfile);
strcat(stat3, suff3);
FILE* out_stat1 = fopen(stat1, "w");
FILE* out_stat2 = fopen(stat2, "w");
FILE* out_stat3 = fopen(stat3, "w");
if ( (out_stat1==NULL)|| (out_stat2==NULL) || (out_stat3==NULL) )
{
die("Unable to open %s or %s or %s to write to - permissions issue?\n", stat1, stat2, stat3);
}
SeqFile *sf = seq_file_open(fastq_file);
if(sf == NULL)
{
// Error opening file
fprintf(stderr, "Error: cannot read seq file '%s'\n", fastq_file);
exit(EXIT_FAILURE);
}
char is_fastq = seq_has_quality_scores(sf);
if (is_fastq==0)
{
die("Error correction is only meant to work on FASTQ and this file: %s is not\n", fastq_file);
}
StrBuf* buf_seq = strbuf_new();
StrBuf* buf_qual = strbuf_new();
StrBuf* working_buf=strbuf_new();
dBNodeEc* last_node_in_read;
Orientation last_or_in_read;
int num_original_reads=0, num_final_reads=0, num_corrected_reads=0, num_discarded_reads=0;
int num_discarded_undefined = 0;
int num_discarded_uncorrectable = 0;
int num_discarded_short_read = 0;
int num_print_uncorrected_lowqual = 0;
int num_print_uncorrected_goodqual = 0;
int num_print_corrected = 0;
StrBuf* buf_dashes = strbuf_create("---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------");
while(seq_next_read(sf))
{
int count_corrected_bases=0;
//NOTE - uses modified version fo Isaacs code - new func
seq_read_all_bases_and_quals(sf, buf_seq, buf_qual);
StrBuf* buf_seq_debug = strbuf_clone(buf_seq);
StrBuf* buf_seq_origin = strbuf_clone(buf_seq);
StrBuf* buf_seq_fixed = strbuf_clone(buf_dashes);
strbuf_resize(buf_seq_fixed, strbuf_len(buf_seq)+1);
int read_len = seq_get_length(sf);
int num_kmers = read_len-kmer_size+1;
int quality_good[read_len];
set_int_array(quality_good, read_len, 1);
int first_good=0;//index of first kmer in graph
//populate the qual array showing which bases have qual >threshold
//if all quals are high, will Print uncorrected
//else, if all kmers NOT in graph, will discard or print uncorrected depending on policy
//else print corrected.
Orientation strand_first_good_kmer;
ReadCorrectionDecison dec =
get_first_good_kmer_and_populate_qual_array(seq_get_read_name(sf), buf_seq, buf_qual, num_kmers, read_len,
quality_good, quality_cutoff,
&first_good, &strand_first_good_kmer,
db_graph, policy, rev_comp_read_if_on_reverse_strand);
//*** start of local functions
//if going right, keep going to right hand end. if going left, keep going to left hand end
boolean condition(WhichEndOfKmer direction, int pos)
{
if ((direction==Right) && (pos<num_kmers))
{
return true;
}
if ((direction==Left) && (pos>=0))
{
return true;
}
return false;
}
boolean kmer_is_in_graph(char* kmer, dBGraphEc* db_g)
{
BinaryKmer curr_kmer;
if (seq_to_binary_kmer(kmer, kmer_size, &curr_kmer)==NULL)
{
//is an N
return false;
}
BinaryKmer temp_key;
element_ec_get_key(&curr_kmer, kmer_size, &temp_key);
dBNodeEc* node = hash_table_ec_find(&temp_key, db_g);
if (node==NULL)
{
return false;
}
else
{
return true;
}
}
int increment(int i, WhichEndOfKmer direction)
{
if (direction==Right)
{
return i+1;
}
else
{
return i-1;
}
}
char working_str[kmer_size+1];
// start_pos is in kmer units
boolean check_bases_to_end_of_read(int start_pos, ReadCorrectionDecison* decision,
WhichEndOfKmer direction,
int* num_corrected_bases_in_this_read_debug)
{
boolean any_correction_done=false;
if ((start_pos<0) || (start_pos>=num_kmers))
{
return any_correction_done;
}
int pos=start_pos;
int offset=0;
if (direction==Right)
{
offset= kmer_size-1;
}
char local_kmer[kmer_size+1];
local_kmer[kmer_size]='\0';
while ( (*decision==PrintCorrected) && (condition(direction,pos)==true) )
{
strncpy(local_kmer, buf_seq->buff+pos, kmer_size);
if (quality_good[pos+offset]==1)
{
//nothing to do
}
else if (kmer_is_in_graph(local_kmer, db_graph)==true)
{
//nothing to do - don't correct if kmer is in graph
}
else//kmer not in graph and quality bad
{
boolean fixed = fix_end_if_unambiguous(direction, buf_seq, buf_seq_fixed, buf_qual, quality_cutoff, pos,
working_buf, working_str, db_graph);
if ( (policy==DiscardReadIfLowQualBaseUnCorrectable)
&&
(fixed==false) )
{
*decision=DiscardUncorrectable;
}
else if (fixed==true)
{
any_correction_done=true;
count_corrected_bases++;
*num_corrected_bases_in_this_read_debug=*num_corrected_bases_in_this_read_debug+1;
if (offset+pos<bases_modified_count_array_size)
{
posn_modified_count_array[offset+pos]++;
}
else
{
posn_modified_count_array[bases_modified_count_array_size-1]++;
}
}
}
pos = increment(pos, direction);
}
return any_correction_done;
}
