?? sequence.c
字號:
/* find the start of the sequence entry */
for (;;) {
while (fgets(line,MAXLINE+1,fin) != NULL)
if( *line == '{' ) break;
while( !keyword(line,"name") )
if (fgets(line,MAXLINE+1,fin) == NULL) return;
for(i=5;i<=strlen(line);i++) /* DES */
if(line[i] != ' ') break;
strncpy(sname,line+i,MAXNAMES); /* remember entryname */
for(i=0;i<=strlen(sname);i++)
if(sname[i] == ' ') {
sname[i]=EOS;
break;
}
sname[MAXNAMES]=EOS;
rtrim(sname);
blank_to_(sname);
/* look for secondary structure feature table / gap penalty mask */
while(fgets(line,MAXLINE+1,fin) != NULL) {
if (keyword(line,"feature")) {
if (interactive) {
strcpy(title,"Found secondary structure in alignment file: ");
strcat(title,sname);
(*lin2)=prompt_for_yes_no(title,"Use it to set local gap penalties ");
}
else (*lin2) = 'y';
if ((*lin2 != 'n') && (*lin2 != 'N')) {
struct_penalties = SECST;
for (i=0;i<length;i++)
sec_struct_mask[i] = '.';
do {
if(keyword(line,"feature"))
get_rsf_feature(&line[7],length);
fgets(line,MAXLINE+1,fin);
} while( !keyword(line,"sequence") );
}
else {
do {
fgets(line,MAXLINE+1,fin);
} while( !keyword(line,"sequence") );
}
strcpy(ss_name,sname);
}
else if (keyword(line,"sequence"))
break;
if (struct_penalties != NONE) break;
}
}
static void get_gde_ss(sint length)
{
static char title[MAXLINE+1];
static char line[MAXLINE+1];
static char lin2[MAXLINE+1];
static char sname[MAXNAMES+1];
sint i, len, offset = 0;
unsigned char c;
for (;;) {
line[0] = '\0';
/* search for the next comment line */
while(*line != '"')
if (fgets(line,MAXLINE+1,fin) == NULL) return;
/* is it a secondary structure entry? */
if (strncmp(&line[1],"SS_",3) == 0) {
for (i=1;i<=MAXNAMES-3;i++) {
if (line[i+3] == '(' || line[i+3] == '\n')
break;
sname[i-1] = line[i+3];
}
i--;
sname[i]=EOS;
if (sname[i-1] == '(') sscanf(&line[i+3],"%d",&offset);
else offset = 0;
for(i--;i > 0;i--)
if(isspace(sname[i])) {
sname[i]=EOS;
}
else break;
blank_to_(sname);
if (interactive) {
strcpy(title,"Found secondary structure in alignment file: ");
strcat(title,sname);
(*lin2)=prompt_for_yes_no(title,"Use it to set local gap penalties ");
}
else (*lin2) = 'y';
if ((*lin2 != 'n') && (*lin2 != 'N')) {
struct_penalties = SECST;
for (i=0;i<length;i++)
sec_struct_mask[i] = '.';
len = 0;
while(fgets(line,MAXLINE+1,fin)) {
if(*line == '%' || *line == '#' || *line == '"') break;
for(i=offset;i < length;i++) {
c=line[i];
if(c == '\n' || c == EOS)
break; /* EOL */
sec_struct_mask[len++]=c;
}
if (len > length) break;
}
strcpy(ss_name,sname);
}
}
/* or is it a gap penalty mask entry? */
else if (strncmp(&line[1],"GM_",3) == 0) {
for (i=1;i<=MAXNAMES-3;i++) {
if (line[i+3] == '(' || line[i+3] == '\n')
break;
sname[i-1] = line[i+3];
}
i--;
sname[i]=EOS;
if (sname[i-1] == '(') sscanf(&line[i+3],"%d",&offset);
else offset = 0;
for(i--;i > 0;i--)
if(isspace(sname[i])) {
sname[i]=EOS;
}
else break;
blank_to_(sname);
if (interactive) {
strcpy(title,"Found gap penalty mask in alignment file: ");
strcat(title,sname);
(*lin2)=prompt_for_yes_no(title,"Use it to set local gap penalties ");
}
else (*lin2) = 'y';
if ((*lin2 != 'n') && (*lin2 != 'N')) {
struct_penalties = GMASK;
for (i=0;i<length;i++)
gap_penalty_mask[i] = '1';
len = 0;
while(fgets(line,MAXLINE+1,fin)) {
if(*line == '%' || *line == '#' || *line == '"') break;
for(i=offset;i < length;i++) {
c=line[i];
if(c == '\n' || c == EOS)
break; /* EOL */
gap_penalty_mask[len++]=c;
}
if (len > length) break;
}
strcpy(ss_name,sname);
}
}
if (struct_penalties != NONE) break;
}
}
static void get_swiss_feature(char *line, sint len)
{
char c, s, feature[MAXLINE+1];
int i, start_pos, end_pos;
if (sscanf(line,"%s%d%d",feature,&start_pos,&end_pos) != 3) {
return;
}
if (strcmp(feature,"HELIX") == 0) {
c = 'A';
s = '$';
}
else if (strcmp(feature,"STRAND") == 0) {
c = 'B';
s = '%';
}
else
return;
if(start_pos >=len || end_pos>=len) return;
sec_struct_mask[start_pos-1] = s;
for (i=start_pos;i<end_pos-1;i++)
sec_struct_mask[i] = c;
sec_struct_mask[end_pos-1] = s;
}
static void get_rsf_feature(char *line, sint len)
{
char c, s;
char str1[MAXLINE+1],str2[MAXLINE+1],feature[MAXLINE+1];
int i, tmp,start_pos, end_pos;
if (sscanf(line,"%d%d%d%s%s%s",&start_pos,&end_pos,&tmp,str1,str2,feature) != 6) {
return;
}
if (strcmp(feature,"HELIX") == 0) {
c = 'A';
s = '$';
}
else if (strcmp(feature,"STRAND") == 0) {
c = 'B';
s = '%';
}
else
return;
if(start_pos>=len || end_pos >= len) return;
sec_struct_mask[start_pos-1] = s;
for (i=start_pos;i<end_pos-1;i++)
sec_struct_mask[i] = c;
sec_struct_mask[end_pos-1] = s;
}
static void get_swiss_mask(char *line, sint len)
{
int i, value, start_pos, end_pos;
if (sscanf(line,"%d%d%d",&value,&start_pos,&end_pos) != 3) {
return;
}
if (value < 1 || value > 9) return;
if(start_pos>=len || end_pos >= len) return;
for (i=start_pos-1;i<end_pos;i++)
gap_penalty_mask[i] = value+'0';
}
static char * get_seq(char *sname,sint *len,char *tit)
{
static char line[MAXLINE+1];
char *seq = NULL;
sint i, offset = 0;
unsigned char c=EOS;
Boolean got_seq=FALSE;
switch(seqFormat) {
/************************************/
case EMBLSWISS:
while( !linetype(line,"ID") )
if (fgets(line,MAXLINE+1,fin) == NULL) return NULL;
for(i=5;i<=strlen(line);i++) /* DES */
if(line[i] != ' ') break;
strncpy(sname,line+i,MAXNAMES); /* remember entryname */
for(i=0;i<=strlen(sname);i++)
if(sname[i] == ' ') {
sname[i]=EOS;
break;
}
sname[MAXNAMES]=EOS;
rtrim(sname);
blank_to_(sname);
while( !linetype(line,"SQ") )
fgets(line,MAXLINE+1,fin);
*len=0;
while(fgets(line,MAXLINE+1,fin)) {
if(got_seq && blankline(line)) break;
if( strlen(line) > 2 && line[strlen(line)-2]=='.' && line[strlen(line)-3]=='.' )
continue;
if(seq==NULL)
seq=(char *)ckalloc((MAXLINE+2)*sizeof(char));
else
seq=(char *)ckrealloc(seq,((*len)+MAXLINE+2)*sizeof(char));
for(i=0;i<=MAXLINE;i++) {
c=line[i];
if(c == '\n' || c == EOS || c == '/')
break; /* EOL */
c=chartab[c];
if(c) {
got_seq=TRUE;
seq[++(*len)]=c;
}
}
if(c == '/') break;
}
break;
/************************************/
case PIR:
while(*line != '>')
fgets(line,MAXLINE+1,fin);
for(i=4;i<=strlen(line);i++) /* DES */
if(line[i] != ' ') break;
strncpy(sname,line+i,MAXNAMES); /* remember entryname */
sname[MAXNAMES]=EOS;
rtrim(sname);
blank_to_(sname);
fgets(line,MAXLINE+1,fin);
strncpy(tit,line,MAXTITLES);
tit[MAXTITLES]=EOS;
i=strlen(tit);
if(tit[i-1]=='\n') tit[i-1]=EOS;
*len=0;
while(fgets(line,MAXLINE+1,fin)) {
if(seq==NULL)
seq=(char *)ckalloc((MAXLINE+2)*sizeof(char));
else
seq=(char *)ckrealloc(seq,((*len)+MAXLINE+2)*sizeof(char));
for(i=0;i<=MAXLINE;i++) {
c=line[i];
if(c == '\n' || c == EOS || c == '*')
break; /* EOL */
c=chartab[c];
if(c) seq[++(*len)]=c;
}
if(c == '*') break;
}
break;
/***********************************************/
case PEARSON:
while(*line != '>')
fgets(line,MAXLINE+1,fin);
for(i=1;i<=strlen(line);i++) /* DES */
if(line[i] != ' ') break;
strncpy(sname,line+i,MAXNAMES); /* remember entryname */
for(i=1;i<=strlen(sname);i++) /* DES */
if(sname[i] == ' ') break;
sname[i]=EOS;
rtrim(sname);
blank_to_(sname);
*tit=EOS;
*len=0;
while(fgets(line,MAXLINE+1,fin)) {
if(seq==NULL)
seq=(char *)ckalloc((MAXLINE+2)*sizeof(char));
else
seq=(char *)ckrealloc(seq,((*len)+MAXLINE+2)*sizeof(char));
for(i=0;i<=MAXLINE;i++) {
c=line[i];
if(c == '\n' || c == EOS || c == '>')
break; /* EOL */
c=chartab[c];
if(c) seq[++(*len)]=c;
}
if(c == '>') break;
}
break;
/**********************************************/
case GDE:
if (dnaflag) {
while(*line != '#')
fgets(line,MAXLINE+1,fin);
}
else {
while(*line != '%')
fgets(line,MAXLINE+1,fin);
}
for (i=1;i<=MAXNAMES;i++) {
if (line[i] == '(' || line[i] == '\n')
break;
sname[i-1] = line[i];
}
i--;
sname[i]=EOS;
if (sname[i-1] == '(') sscanf(&line[i],"%d",&offset);
else offset = 0;
for(i--;i > 0;i--)
if(isspace(sname[i])) {
sname[i]=EOS;
}
else break;
blank_to_(sname);
*tit=EOS;
*len=0;
for (i=0;i<offset;i++) seq[++(*len)] = '-';
while(fgets(line,MAXLINE+1,fin)) {
if(*line == '%' || *line == '#' || *line == '"') break;
if(seq==NULL)
seq=(char *)ckalloc((MAXLINE+2)*sizeof(char));
else
seq=(char *)ckrealloc(seq,((*len)+MAXLINE+2)*sizeof(char));
for(i=0;i<=MAXLINE;i++) {
c=line[i];
if(c == '\n' || c == EOS)
break; /* EOL */
c=chartab[c];
if(c) seq[++(*len)]=c;
}
}
break;
/***********************************************/
case RSF:
while(*line != '{')
if (fgets(line,MAXLINE+1,fin) == NULL) return NULL;
while( !keyword(line,"name") )
if (fgets(line,MAXLINE+1,fin) == NULL) return NULL;
for(i=5;i<=strlen(line);i++) /* DES */
if(line[i] != ' ') break;
strncpy(sname,line+i,MAXNAMES); /* remember entryname */
for(i=0;i<=strlen(sname);i++)
if(sname[i] == ' ') {
sname[i]=EOS;
break;
}
sname[MAXNAMES]=EOS;
rtrim(sname);
blank_to_(sname);
while( !keyword(line,"sequence") )
if (fgets(line,MAXLINE+1,fin) == NULL) return NULL;
*len=0;
while(fgets(line,MAXLINE+1,fin)) {
if(seq==NULL)
seq=(char *)ckalloc((MAXLINE+2)*sizeof(char));
else
seq=(char *)ckrealloc(seq,((*len)+MAXLINE+2)*sizeof(char));
for(i=0;i<=MAXLINE;i++) {
c=line[i];
if(c == EOS || c == '}')
break; /* EOL */
if( c=='.')
seq[++(*len)]='-';
c=chartab[c];
if(c) seq[++(*len)]=c;
}
if(c == '}') break;
}
break;
/***********************************************/
}
seq[*len+1]=EOS;
return seq;
}
sint readseqs(sint first_seq) /*first_seq is the #no. of the first seq. to read */
{
?? 快捷鍵說明
復制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -