int putsPivot( int * tab, int begin, int end ) {
int i;
int p = begin;
for( i = p+1; i < end; ++i )
if( tab[i] < tab[p] ) {
swapi( &tab[i], &tab[p] );
swapi( &tab[i], &tab[p+1] );
++p;
}
return p;
}
void bubble_sort( int *t, int size ) {
int i,j;
for(i=0; i<size-1; ++i)
for(j=size-1; j>i; j--)
if(t[j]<t[j-1])
swapi(&t[j],&t[j-1]);
}
void swim(int a[], int k)
{
while (k > 1 && lessi(a, k / 2, k)) {
swapi(a, k, k / 2);
k /= 2;
}
}
int nbt_write_int(nbt_file *nbt, int32_t *val)
{
int32_t temp = *val;
if (get_endianness() == L_ENDIAN)
swapi((uint32_t *)&temp);
return gzwrite(nbt->fp, &temp, sizeof(int32_t));
}
int nbt_read_byte_array(nbt_file *nbt, unsigned char **out)
{
int32_t len;
gzread(nbt->fp, &len, sizeof(len));
if (get_endianness() == L_ENDIAN)
swapi((uint32_t *)&len);
*out = malloc(len);
gzread(nbt->fp, *out, len);
return len;
}
int nbt_read_int(nbt_file *nbt, int32_t **out)
{
int32_t t;
gzread(nbt->fp, &t, sizeof(t));
if (get_endianness() == L_ENDIAN)
swapi((uint32_t *)&t);
*out = malloc(sizeof(int32_t));
memcpy(*out, &t, sizeof(int32_t));
return 0;
}
int del_max(int a[])
{
// get the max
int max = a[1];
// swap first (root) with last elem
swapi(a, 1, N);
a[N--] = DEFAULT_INT;
// sink the new root to the correct position
sink(a, 1);
return max;
}
int partition(int array[], int lo, int hi)
{
int i = lo;
int j = hi + 1;
while (1) {
while (lessi(array, ++i, lo))
if (i == hi)
break;
while (lessi(array, lo, --j))
if (j == lo)
break;
if (i >= j)
break;
swapi(array, i, j);
}
swapi(array, lo, j);
return j;
}
void sink(int a[], int k)
{
while (2 * k <= N) {
int j = 2 * k;
// children of k are 2k and 2k + 1
if (j < N && lessi(a, j, j + 1))
++j;
if (lessi(a, j, k))
break;
swapi(a, k, j);
k = j;
}
}
static void switchtonextlevel(void)
{
struct initrec** inittab = cfg->inittab;
struct initrec** pp;
struct initrec* p;
/* One we're here, reset pid for r-type entries, to run them when
entering another runlevel with shouldberunning(p) true. */
for(pp = inittab; (p = *pp); pp++)
if(!shouldberunning(p) && once(p) && (p->pid < 0))
p->pid = 0;
if(slippery(nextlevel) && !slippery(currlevel)) {
/* nextlevel is slippery, turn back to currlevel */
swapi(&currlevel, &nextlevel);
/* We've got to make sure pollfds will return immediately. */
timetowait = 0;
} else {
currlevel = nextlevel;
}
}
int32_t nbt_read_list(nbt_file *nbt, char *type_out, void ***target)
{
char type;
int32_t len;
int i;
gzread(nbt->fp, &type, 1);
*type_out = type;
gzread(nbt->fp, &len, sizeof(len));
if (get_endianness() == L_ENDIAN)
swapi((uint32_t *)&len);
*target = malloc(len * sizeof(void *));
for (i = 0; i < len; ++i)
nbt_read(nbt, type, &((*target)[i]));
return len;
}
int main(int argc, char* argv[]) {
char cpsfilename[MAXBUFFLENGTH];
char chainfilename[MAXBUFFLENGTH];
char outfilename[MAXBUFFLENGTH]="";
int marginlength = 0;
int MAXREC;
char c;
char *pc;
int x;
char buff[MAXBUFFLENGTH+1];
char aux[MAXBUFFLENGTH+1];
long score;
int start1,end1,len1,start2,end2,len2;
char strand1, strand2, chr1[MAXBUFFLENGTH], chr2[MAXBUFFLENGTH];
int *size, *dq, *dt;
int a,b,k,i,j,s,m;
int *position;
char *strand;
int *ids;
int *idg;
char *type;
int chridx[MAXCHR+1];
int chroff[MAXCHR+1];
char resstr;
int rescrd;
if(argc==1) {
fprintf(stderr,"Finds matches of the given set of sites (CPS file) in the BLASTZ chain alignment (CHAIN file)\n");
fprintf(stderr,"Last update by (dp) on Sep 21, 2011\n");
fprintf(stderr,"Keys:\n -i CPS file (remember to sort by position in ascending order)\n -d CHAIN alignment file\n -o output file\n");
fprintf(stderr," -m margin length [0]\n -v suppress verbose output [NO]\n");
exit(1);
}
timestamp_set();
for(i=1;i<argc;i++) {
pc = argv[i];
if(*pc != '-') continue;
if(*(pc+1) == 'i') {
sscanf(argv[++i], "%s", &cpsfilename[0]);
}
if(*(pc+1) == 'd') {
sscanf(argv[++i], "%s", &chainfilename[0]);
}
if(*(pc+1) == 'o') {
sscanf(argv[++i], "%s", &outfilename[0]);
}
if(*(pc+1) == 'm') {
sscanf(argv[++i], "%i", &marginlength);
}
if(*(pc+1) == 'v') {
verbose=0;
}
}
if(outfilename[0]==0) {
fprintf(stderr,"No output file privided, exiting\n");
exit(1);
}
outfile = fopen(outfilename,"w");
if(outfile == NULL) {
fprintf(stderr,"Can't open output file, exiting\n");
exit(1);
}
for(i=0;i<MAXCHR;i++) {
chridx[i]=chroff[i]=0;
}
MAXREC = 0;
cpsfile= fopen(cpsfilename,"r");
if(cpsfile==NULL) {
fprintf(stderr,"Can't access CPS file. Exiting\n");
exit(1);
}
if(verbose) fprintf(stderr,"Reading CPS input pass 1");
while(!feof(cpsfile)) {
buff[0]=0;
fgets(buff,MAXBUFFLENGTH,cpsfile);
if(strlen(buff)<2) break;
sscanf(buff,"%s" , aux);
chridx[assign_code(aux)]++;
MAXREC++;
}
fclose(cpsfile);
for(s=i=0;i<MAXCHR;i++) {
x = chridx[i];
chridx[i] =s;
s+=x;
}
chridx[i] = s;
position = (int*) malloc(sizeof(int)*(s+4));
strand = (char*) malloc(sizeof(char)*(s+4));
type = (char*) malloc(sizeof(char)*(s+4));
ids = (int*) malloc(sizeof(int)*(s+4));
idg = (int*) malloc(sizeof(int)*(s+4));
if(position==NULL || strand==NULL || type==NULL || ids==NULL || idg==NULL) {
fprintf(stderr,"Not enough memory. Terminated\n");
exit(1);
}
cpsfile= fopen(cpsfilename,"r");
if(verbose) fprintf(stderr,", records = %i\nReading CPS input pass 2",MAXREC);
while(!feof(cpsfile)) {
buff[0]=0;
fgets(buff,MAXBUFFLENGTH,cpsfile);
if(strlen(buff)<2) break;
sscanf(buff,"%s" , aux);
i = assign_code(aux);
m = chridx[i]+chroff[i];
sscanf(buff,"%*s %i %c %i %i %c" , position+m, strand+m, idg+m, ids+m, type+m);
chroff[i]++;
}
fclose(cpsfile);
if(verbose) fprintf(stderr,"\nSorting segments");
/*
for(i=0;i<MAXCHR;i++) {
quickSort_ic(position,strand,chridx[i],chridx[i+1]-1);
}
*/
for(i=0;i<MAXCHR;i++) {
k=1;
while(k) {
k=0;
for(j=chridx[i];j<chridx[i+1]-1;j++) {
if(position[j]>position[j+1]) {
k=1;
swapi(position+j,position+j+1);
swapc(strand+j,strand+j+1);
swapc(type+j,type+j+1);
swapi(ids+j,ids+j+1);
swapi(idg+j,idg+j+1);
}
}
}
}
if(verbose) fprintf(stderr," done\nProcessing chains");
/**********************************************************************************************/
size = (int*) malloc(sizeof(int)*MAXALN);
dq = (int*) malloc(sizeof(int)*MAXALN);
dt = (int*) malloc(sizeof(int)*MAXALN);
if(size ==0 || dq ==0 || dt==0) {
fprintf(stderr,"Not enough memory for such long chains. Terminated\n");
exit(1);
}
/**********************************************************************************************/
chainfile = fopen(chainfilename,"r");
while(!feof(chainfile)) {
buff[0]=0;
fgets(buff,MAXBUFFLENGTH,chainfile);
if(strlen(buff)<2) break;
buff[5]=0;
if(strcmp(buff,"chain")==0) {
sscanf(buff+6,"%li %s %i %c %i %i %s %i %c %i %i",&score, &chr1[0], &len1, &strand1, &start1, &end1, &chr2[0], &len2, &strand2, &start2, &end2);
k=0;
while(!feof(chainfile)) {
buff[0]=0;
fgets(buff,MAXBUFFLENGTH,chainfile);
if(strlen(buff)<2) break;
sscanf(buff,"%i %i %i",&size[k],&dt[k],&dq[k]);
k++;
if(k>MAXALN) {
fprintf(stderr,"Chain length exceeded. Terminating");
exit(1);
}
}
x = get_chr_code(chr1);
if(x<0) continue;
a=start1;b=start2;
j=0;
for(i=chridx[x];i<chridx[x+1] && position[i]<start1;i++);
for(;i<chridx[x+1]&& position[i]<end1;i++) {
while(position[i]>a+size[j]+dt[j] && j<k){
a+=size[j]+dt[j];
b+=size[j]+dq[j];
j++;
}
if(j>=k) break;
if(position[i]-a > marginlength && a+size[j]-position[i] >= marginlength) {
if(strand1==strand2) {
resstr = strand[i];
rescrd = position[i] - a + b;
}
else {
resstr = (strand[i]=='+') ? '-' : '+';
rescrd = len2 - (position[i] - a + b - 1) ;
}
fprintf(outfile,"%s\t%i\t%c\t%s\t%i\t%c\t%i\t%i\t%c\t%li\n",chr1, position[i], strand[i],chr2,rescrd, resstr, idg[i], ids[i],type[i],score);
}
}
}
}
if(verbose) fprintf(stderr," done\n");
fclose(chainfile);
fclose(outfile);
timestamp_report();
exit(0);
}
double minver(double a[], int l, int m, double eps)
{
int i, iw, j, k, *p, r, s, t, u, v, *work;
double api, pivot, *q, *q1, w, w1, wmax;
if(m < 2 || l < m || eps <= 0.)
{
fprintf(stderr, "Error : Illegal parameter in minver()\n");
return 0.;
}
work = (int *)malloc(m * sizeof(int));
if(work == NULL)
{
fprintf(stderr, "Error : Out of memory in minver()\n");
return 0.;
}
w1 = 1.;
for(i = 0, p = work; i < m; i++) *p++ = i;
for(k = 0, u = 0; k < m; k++, u += l)
{
wmax = 0.;
for(i = k; i < m; i++)
{
w = fabs(*(a + i * l + k));
if(w > wmax)
{
wmax = w;
r = i;
}
}
api = fabs(pivot = *(a + r * l + k));
if(api < eps)
{
fprintf(stderr, "Error : api < eps in minver()\n");
free((char *)work);
return w1;
}
w1 *= pivot;
v = r * l;
if(r != k)
{
w1 = -w1;
swapi(work + k, work + r);
for(j = 0, q = a + u, q1 = a + v; j < m; j++) swapd(q++, q1++);
}
for(i = 0, q = a + u; i < m; i++) *q++ /= pivot;
for(i = 0, v = 0; i < m; i++, v += l)
{
if(i != k)
{
s = v + k;
w = *(a + s);
if(w != 0.)
{
for(j = 0, q = a + u, q1 = a + v; j < m; j++, q++, q1++)
if (j != k) *q1 -= w * *q;
*(a + s) = - w / pivot;
}
}
}
*(a + u + k) = 1. / pivot;
}
for(i = 0; i < m; i++)
{
for(;;)
{
k = *(work + i);
if(k == i) break;
swapi(work + k, work + i);
for(j = 0, u = 0; j < m; j++, u += l) swapd(a + u + i, a + u + k);
}
}
free((char *)work);
return w1;
}
int main(int argc, char* argv[]) {
char cps_file_name[MAXBUFFLENGTH];
char chain_file_name[MAXBUFFLENGTH];
char out_file_name[MAXBUFFLENGTH]="";
int marginlength = 0;
char buff[MAXBUFFLENGTH+1];
char aux[MAXBUFFLENGTH+1];
long score;
int start1,end1,len1,start2,end2,len2;
char strand1, strand2, chr1[MAXBUFFLENGTH], chr2[MAXBUFFLENGTH];
int *size, *dq, *dt;
int a,b,k,i,j,s,m,x;
int *position;
int *strand;
int *idg;
int *ids;
char *type;
char c;
int chridx[MAXCHR+1];
int chroff[MAXCHR+1];
char resstr;
int rescrd;
long chain_id;
if(argc==1) {
fprintf(stderr,"This utility does liftOver of coordinates (cps) by using chain alignment\n");
fprintf(stderr,"Gene information is included in the output\n");
fprintf(stderr,"Last update by Dmitri Pervouchine ([email protected]) on Mar 22, 2013\n");
fprintf(stderr,"Usage: %s -in <cps_file> -chain <chain_alignment_file> [-margin <length>] [-quiet]\n", argv[0]);
fprintf(stderr," -in cps6, i.e. chr1/position1/strand1/gene/site/type tab-delimited file, strand is +/-\n");
fprintf(stderr," -chain UCSC chain alignment file, species1=>2\n");
fprintf(stderr," -out <output_file> [default=stdout]\n");
fprintf(stderr," -margin margin length [default=0]\n -quiet suppress verbose output [default=NO]\n");
fprintf(stderr,"NOTE: Input has to be sorted by position!\n");
fprintf(stderr,"Output format cps3+cps6: chr1/position1/strand1/chr2/position2/strand2/gene/site/type/score\n");
exit(1);
}
timestamp_set();
for(i=1;i<argc;i++) {
if(strcmp(argv[i],"-in")==0) {
sscanf(argv[++i], "%s", &cps_file_name[0]);
}
if(strcmp(argv[i],"-chain")==0) {
sscanf(argv[++i], "%s", &chain_file_name[0]);
}
if(strcmp(argv[i],"-out")==0) {
sscanf(argv[++i], "%s", &out_file_name[0]);
}
if(strcmp(argv[i],"-margin")==0) {
sscanf(argv[++i], "%i", &marginlength);
}
if(strcmp(argv[i],"-quiet")==0) {
verbose=0;
}
}
if(out_file_name[0]==0) {
fprintf(stderr,"[WARNING: output file not specified, redirect to stdout]\n");
out_file = stdout;
}
else {
out_file = fopen(out_file_name,"w");
if(out_file == NULL) {
fprintf(stderr,"[ERROR: output file %s cannot be opened for writing, exiting]\n", out_file_name);
exit(1);
}
if(verbose) fprintf(stderr,"[>%s]\n",out_file_name);
}
cps_file= fopen(cps_file_name,"r");
if(cps_file==NULL) {
fprintf(stderr,"[ERROR: cannot access %s, exiting]\n", cps_file_name);
exit(1);
}
for(i=0;i<MAXCHR;i++) chridx[i] = chroff[i] = 0;
if(verbose) fprintf(stderr,"[<%s, pass 1",cps_file_name);
while(fgets(buff,MAXBUFFLENGTH,cps_file)) {
if(strlen(buff)<2) break;
sscanf(buff,"%s" , aux);
chridx[assign_code(aux)]++;
}
if(verbose) fprintf(stderr,"]\n");
for(s=i=0;i<MAXCHR;i++) {
x = chridx[i];
chridx[i] =s;
s+=x;
}
chridx[i] = s;
position = (int*) malloc(sizeof(int)*(s + ARRAY_MARGIN));
strand = (int*) malloc(sizeof(int)*(s + ARRAY_MARGIN));
ids = (int*) malloc(sizeof(int)*(s + ARRAY_MARGIN));
idg = (int*) malloc(sizeof(int)*(s + ARRAY_MARGIN));
type = (char*) malloc(sizeof(char)*(s + ARRAY_MARGIN));
if(position==NULL || strand==NULL || type==NULL || ids==NULL || idg==NULL) {
fprintf(stderr,"[ERROR: failed to create index tables, exiting]\n");
exit(1);
}
fseek (cps_file, 0, SEEK_SET);
if(verbose) fprintf(stderr,"[<%s, pass 2", cps_file_name);
while(fgets(buff,MAXBUFFLENGTH,cps_file)) {
if(strlen(buff)<2) break;
sscanf(buff,"%s" , aux);
i = assign_code(aux);
m = chridx[i]+chroff[i];
sscanf(buff,"%*s %i %c %i %i %c" , &position[m], &c,&idg[m],&ids[m],&type[m]);
strand[m] = strand_c2i(c);
chroff[i]++;
}
fclose(cps_file);
if(verbose) fprintf(stderr,"]\n");
if(verbose) fprintf(stderr,"[Sort by position (if not done before)");
for(i=0;i<MAXCHR;i++) {
k=1;
while(k) {
k=0;
for(j=chridx[i];j<chridx[i+1]-1;j++) {
if(position[j]>position[j+1]) {
k=1;
swapi(position+j,position+j+1);
swapi(strand+j,strand+j+1);
swapi(idg+j,idg+j+1);
swapi(ids+j,ids+j+1);
swapc(type+j,type+j+1);
}
}
}
}
if(verbose) fprintf(stderr,"]\n");
/**********************************************************************************************/
size = (int*) malloc(sizeof(int)*(MAXALN + ARRAY_MARGIN));
dq = (int*) malloc(sizeof(int)*(MAXALN + ARRAY_MARGIN));
dt = (int*) malloc(sizeof(int)*(MAXALN + ARRAY_MARGIN));
if(size ==0 || dq ==0 || dt==0) {
fprintf(stderr,"[ERROR: not enough memory for chains, exiting]\n");
exit(1);
}
/**********************************************************************************************/
chain_file = fopen(chain_file_name,"r");
if(chain_file==NULL) {
fprintf(stderr,"[ERROR: cannot access %s, exiting]\n", chain_file_name);
exit(1);
}
fseek(chain_file, 0, SEEK_END);
unsigned int last_pos = ftell(chain_file);
fseek(chain_file, 0, SEEK_SET);
while(fgets(buff,MAXBUFFLENGTH,chain_file)) {
if(strlen(buff)<2) break;
buff[5]=0;
if(strcmp(buff,"chain")==0) {
sscanf(buff+6,"%li %s %i %c %i %i %s %i %c %i %i %li",&score, &chr1[0], &len1, &strand1, &start1, &end1, &chr2[0], &len2, &strand2, &start2, &end2, &chain_id);
k=0;
while(fgets(buff,MAXBUFFLENGTH,chain_file)) {
if(strlen(buff)<2) break;
progressbar(ftell(chain_file), last_pos-1, (char*)"Processing ", verbose);
sscanf(buff,"%i %i %i",size + k, dt + k, dq + k);
k++;
if(k>=MAXALN) {
fprintf(stderr,"[ERROR: chain too long, exiting]\n");
exit(1);
}
}
x = get_chr_code(chr1);
if(x<0) continue;
a=start1;b=start2;
j=0;
for(i=chridx[x];i<chridx[x+1] && position[i]<start1;i++);
for(;i<chridx[x+1]&& position[i]<end1;i++) {
while(position[i]>a+size[j]+dt[j] && j<k){
a+=size[j]+dt[j];
b+=size[j]+dq[j];
j++;
}
if(j>=k) break;
if(position[i]-a > marginlength && a+size[j]-position[i] >= marginlength) {
if(strand1==strand2) {
resstr = strand[i];
rescrd = position[i] - a + b;
}
else {
resstr = -strand[i];
rescrd = len2 - (position[i] - a + b - 1) ;
}
fprintf(out_file,"%s\t%i\t%c\t%s\t%i\t%c\t%i\t%i\t%c\t%li\n",chr1, position[i], strand_i2c(strand[i]), chr2, rescrd, strand_i2c(resstr), idg[i], ids[i],type[i],score);
}
}
}
}
fclose(chain_file);
fclose(out_file);
timestamp_report();
free(size);
free(dq);
free(dt);
free(position);
free(strand);
exit(0);
}
