/// Reads a SNP list in .gff and puts the SNPs as IUPAC bases into the chromosome; superceded by incorporate_all_gff().
void TChromosome::incorporate_known_snps_gff ( string filename ) {
if ( DEBUGGING ) { fprintf ( stdout , "Reading SNPS into %s from %s ... " , name.c_str() , filename.c_str() ) ; fflush(stdout); }
FILE * file = fopen ( filename.c_str() , "r" ) ;
int snpcnt = 0 ;
char dummy[READ_CHAR_BUFFER] , *c1 , *c2 ;
while ( !feof(file) ) {
vector <string> parts , parts2 ;
*dummy = 0 ;
fgets ( dummy , READ_CHAR_BUFFER , file ) ;
Tokenize ( dummy , parts , "\t" ) ;
if ( parts.size() < 7 ) continue ; // These is not the data you are looking for
if ( parts[0] != name ) continue ; // Wrong chromosome
if ( parts[2] != "SNP" ) continue ; // Not a SNP
if ( parts[3] != parts[4] ) continue ; // Not a single point
if ( parts[5] != "+" ) continue ; // Wrong strand (?)
int pos = atoi ( parts[3].c_str() ) ;
char merge = sequence[pos-1] ;
Tokenize ( parts[6].c_str() , parts2 , "\"\" \"\"" ) ;
bool found_reference = false ;
for ( int b = 0 ; b < parts2.size() ; b++ ) {
int l = parts2[b].length() ;
if ( l < 2 ) continue ;
if ( parts2[b][l-2] != ':' ) continue ;
char c = parts2[b][l-1] ;
if ( c == sequence[pos-1] ) found_reference = true ;
merge = MERGE_IUPAC(merge,c) ;
}
if ( !found_reference ) { // Strange; maybe original sequence contains "N"
// We should count this!
continue ;
}
sequence[pos-1] = merge ;
snpcnt++ ;
}
fclose ( file ) ;
if ( DEBUGGING ) { fprintf ( stdout , "incorporated %d SNPs.\n" , snpcnt ) ; fflush(stdout); }
}
/// Reads a SNP list in simple format and puts the SNPs as IUPAC bases into the chromosome.
void TChromosome::incorporate_known_snps ( string filename ) {
if ( DEBUGGING ) { fprintf ( stdout , "Reading from %s ... " , filename.c_str() ) ; fflush(stdout); }
FILE * file = fopen ( filename.c_str() , "r" ) ;
char dummy[READ_CHAR_BUFFER] , *c1 , *c2 ;
int snpcnt = 0 ;
while ( !feof(file) ) {
vector <string> parts ;
*dummy = 0 ;
fgets ( dummy , READ_CHAR_BUFFER , file ) ;
Tokenize ( dummy , parts , "\t" ) ;
if ( parts.size() != 4 ) continue ;
if ( name != parts[0] ) continue ;
int pos = atoi ( parts[1].c_str() ) ;
uchar orig = parts[2][0] ;
uchar snp = parts[3][0] ;
sequence[pos-1] = MERGE_IUPAC(orig,snp) ;
snpcnt++ ;
}
fclose ( file ) ;
if ( DEBUGGING ) { fprintf ( stdout , "incorporated %d SNPs.\n" , snpcnt ) ; fflush(stdout); }
}
void TAlignmentOutput::show_pileup ( FILE *pileup , bool snps_only ) {
int search_snps = 0 , found_snps = 0 , bogus = 0 ;
uint base , l , cnt , p , lastns ;
uchar common ;
char *s2 = new char[MAXLINES+5] ;
char *s3 = new char[MAXLINES+5] ;
char *t , *u ;
uint count[256] ;
for ( base = 0 ; base < chr->sequence.length() ; base++ ) {
if ( snps_only && !isIUPAC[chr->sequence[base]] ) continue ;
if ( isIUPAC[chr->sequence[base]] ) search_snps++ ;
count['A']=count['C']=count['G']=count['T']=0 ;
cnt = 0 ;
common = ' ' ;
lastns = 0 ;
p = twotoone(0,base) ;
t = s2 ;
u = s3 ;
char *c = align + p ;
char *q = qalign + p ;
for ( l = 0 ; l < MAXLINES ; l++ , c++ , q++ ) {
if ( *c && *c != ' ' ) {
lastns = l ;
cnt++ ;
count[*c]++ ;
common = MERGE_IUPAC ( *c , common ) ;
if ( *q < MINQUAL + 33 ) *t++ = to_lc[*c] ;
else *t++ = *c ;
*u++ = *q ;
} else {
*t++ = ' ' ;
*u++ = ' ' ;
}
}
s2[lastns+1] = 0 ;
s3[lastns+1] = 0 ;
bool confirmed_snp = false ;
uint n = 0 ;
if ( count['A'] >= MINOCCUR ) n++ ;
if ( count['C'] >= MINOCCUR ) n++ ;
if ( count['G'] >= MINOCCUR ) n++ ;
if ( count['T'] >= MINOCCUR ) n++ ;
if ( n > 1 ) {
confirmed_snp = true ;
found_snps++ ;
}
/*
if ( common != ' ' && isIUPAC[common] ) {
confirmed_snp = true ;
found_snps++ ;
}
*/
if ( common != ' ' && isIUPAC[common] && !isIUPAC[chr->sequence[base]] ) bogus++ ;
if ( snps_only && common == ' ' ) continue ;
char mark = confirmed_snp ? '*' : ' ' ;
string ref ;
if ( !chr->original_sequence.empty() ) {
ref += chr->original_sequence[base] ;
ref += "\t" ;
}
ref += chr->sequence[base] ;
fprintf ( pileup , "%s\t%s\t%d\t%c%c\t%d\t" , chr->name.c_str() , ref.c_str() , base+1 , common , mark , cnt ) ;
fprintf ( pileup , "%s\t%s\n" , s2 , s3 ) ;
}
}
