cutsites* sliding_window (kseq_t *fqrec, int qualtype, int length_threshold, int qual_threshold, int no_fiveprime, int discard_n) {
int window_size = (int) (0.1 * fqrec->seq.l);
int i,j;
int window_start=0;
int window_total=0;
int three_prime_cut = fqrec->seq.l;
int five_prime_cut = 0;
int found_five_prime = 0;
double window_avg;
cutsites* retvals;
/* If the sequence contains an "N" then discard if the option has been selected */
/* Also discard if the length of the sequence is less than the length threshold */
if ((discard_n && (strstr(fqrec->seq.s, "N") || strstr(fqrec->seq.s, "n"))) ||
(fqrec->seq.l < length_threshold)) {
retvals = (cutsites*) malloc (sizeof(cutsites));
retvals->three_prime_cut = -1;
retvals->five_prime_cut = -1;
return (retvals);
}
/* if the seq length is less then 10bp, */
/* then make the window size the length of the seq */
if (window_size == 0) window_size = fqrec->seq.l;
for (i=0; i<window_size; i++) {
window_total += get_quality_num (fqrec->qual.s[i], qualtype, fqrec, i);
}
for (i=0; i <= fqrec->qual.l - window_size; i++) {
window_avg = (double)window_total / (double)window_size;
/* If it is the first window, and the qual average is already above
the threshold, then we have already found the five prime cut at pos 0 */
if (i==0 && window_avg >= qual_threshold) {found_five_prime = 1;}
/* Finding the 5' cutoff */
/* Find when the average quality in the window goes above the threshold starting from the 5' end */
if (no_fiveprime == 0 && found_five_prime == 0 && window_avg >= qual_threshold) {
/* at what point in the window does the quality go above the threshold? */
for (j=window_start; j<window_start+window_size; j++) {
if (get_quality_num (fqrec->qual.s[j], qualtype, fqrec, j) >= qual_threshold) {
five_prime_cut = j;
break;
}
}
found_five_prime = 1;
}
/* Finding the 3' cutoff */
/* if the average quality in the window is less than the threshold */
/* or if the window is the last window in the read */
if ((window_avg < qual_threshold ||
window_start+window_size > fqrec->qual.l) && found_five_prime == 1) {
/* at what point in the window does the quality dip below the threshold? */
for (j=window_start; j<window_start+window_size; j++) {
if (get_quality_num (fqrec->qual.s[j], qualtype, fqrec, j) < qual_threshold) {
three_prime_cut = j;
/* if cutting length is less than threshold then return -1 for both */
/* to indicate that the read should be discarded */
if (three_prime_cut - five_prime_cut < length_threshold) {
three_prime_cut = -1;
five_prime_cut = -1;
}
break;
}
}
break;
}
/* instead of sliding the window, subtract the first qual and add the next qual */
window_total -= get_quality_num (fqrec->qual.s[window_start], qualtype, fqrec, window_start);
if (window_start+window_size < fqrec->qual.l) {
window_total += get_quality_num (fqrec->qual.s[window_start+window_size], qualtype, fqrec, window_start+window_size);
}
window_start++;
}
/* If you never find a five prime cut site, then discard whole read */
if (found_five_prime == 0) {
three_prime_cut = -1;
five_prime_cut = -1;
}
retvals = (cutsites*) malloc (sizeof(cutsites));
retvals->three_prime_cut = three_prime_cut;
retvals->five_prime_cut = five_prime_cut;
return (retvals);
}
cutsites* sliding_window (kseq_t *fqrec, int qualtype, int length_threshold, int qual_threshold, int no_fiveprime, int trunc_n, int trunc_size, int debug) {
if (trunc_size >= 0 && trunc_size < length_threshold) {
trunc_size = length_threshold;
}
int window_size = (int) (0.1 * fqrec->seq.l);
int i,j;
int window_start=0;
int window_total=0;
int three_prime_cut = fqrec->seq.l;
int five_prime_cut = 0;
int found_five_prime = 0;
double window_avg;
cutsites* retvals;
char *npos;
/* discard if the length of the sequence is less than the length threshold */
if (fqrec->seq.l < length_threshold) {
retvals = (cutsites*) malloc (sizeof(cutsites));
retvals->three_prime_cut = -1;
retvals->five_prime_cut = -1;
return (retvals);
}
/* if the seq length is less then 10bp, */
/* then make the window size the length of the seq */
if (window_size == 0) window_size = fqrec->seq.l;
for (i=0; i<window_size; i++) {
window_total += get_quality_num (fqrec->qual.s[i], qualtype, fqrec, i);
}
for (i=0; i <= fqrec->qual.l - window_size; i++) {
window_avg = (double)window_total / (double)window_size;
if (debug) printf ("no_fiveprime: %d, found 5prime: %d, window_avg: %f\n", no_fiveprime, found_five_prime, window_avg);
/* Finding the 5' cutoff */
/* Find when the average quality in the window goes above the threshold starting from the 5' end */
if (no_fiveprime == 0 && found_five_prime == 0 && window_avg >= qual_threshold) {
if (debug) printf ("inside 5-prime cut\n");
/* at what point in the window does the quality go above the threshold? */
for (j=window_start; j<window_start+window_size; j++) {
if (get_quality_num (fqrec->qual.s[j], qualtype, fqrec, j) >= qual_threshold) {
five_prime_cut = j;
break;
}
}
if (debug) printf ("five_prime_cut: %d\n", five_prime_cut);
found_five_prime = 1;
}
/* Finding the 3' cutoff */
/* if the average quality in the window is less than the threshold */
/* or if the window is the last window in the read */
if ((window_avg < qual_threshold ||
window_start+window_size > fqrec->qual.l) && (found_five_prime == 1 || no_fiveprime)) {
/* at what point in the window does the quality dip below the threshold? */
for (j=window_start; j<window_start+window_size; j++) {
if (get_quality_num (fqrec->qual.s[j], qualtype, fqrec, j) < qual_threshold) {
three_prime_cut = j;
break;
}
}
break;
}
/* instead of sliding the window, subtract the first qual and add the next qual */
window_total -= get_quality_num (fqrec->qual.s[window_start], qualtype, fqrec, window_start);
if (window_start+window_size < fqrec->qual.l) {
window_total += get_quality_num (fqrec->qual.s[window_start+window_size], qualtype, fqrec, window_start+window_size);
}
window_start++;
}
/* If truncate N option is selected, and sequence has Ns, then */
/* change 3' cut site to be the base before the first N */
if (trunc_n && ((npos = strstr(fqrec->seq.s, "N")) || (npos = strstr(fqrec->seq.s, "n")))) {
three_prime_cut = npos - fqrec->seq.s;
}
/* if cutting length is less than threshold then return -1 for both */
/* to indicate that the read should be discarded */
/* Also, if you never find a five prime cut site, then discard whole read */
if ((found_five_prime == 0 && !no_fiveprime) || (three_prime_cut - five_prime_cut < length_threshold)) {
three_prime_cut = -1;
five_prime_cut = -1;
if (debug) printf("%s\n", fqrec->name.s);
}
if (debug) printf ("\n\n");
/* if trunc_size is less than read length, remove equal number of bases from each side */
/* TODO: this should acutally remove bases based on quality */
if (trunc_size >= 0) {
int side = 0;
while (trunc_size < three_prime_cut - five_prime_cut) {
if (side == 0) {
five_prime_cut++;
side = 1;
}
else {
three_prime_cut--;
side = 0;
}
}
}
retvals = (cutsites*) malloc (sizeof(cutsites));
retvals->three_prime_cut = three_prime_cut;
retvals->five_prime_cut = five_prime_cut;
return (retvals);
}
