| 1 | #include <stdio.h> |
|---|
| 2 | #include "convert.h" |
|---|
| 3 | #include "global.h" |
|---|
| 4 | |
|---|
| 5 | /* ------------------------------------------------------------------ */ |
|---|
| 6 | /* Function to_gcg(). |
|---|
| 7 | /* Convert from whatever to GCG format. |
|---|
| 8 | */ |
|---|
| 9 | void |
|---|
| 10 | to_gcg(intype, inf) |
|---|
| 11 | int intype; |
|---|
| 12 | char *inf; |
|---|
| 13 | { |
|---|
| 14 | FILE *fopen(), *ifp1, *ifp2, *ifp3, *ofp; |
|---|
| 15 | char temp[TOKENNUM], *eof, line[LINENUM], key[TOKENNUM]; |
|---|
| 16 | char line1[LINENUM], line2[LINENUM], line3[LINENUM], name[LINENUM]; |
|---|
| 17 | char *eof1, *eof2, *eof3; |
|---|
| 18 | char outf[TOKENNUM], *Fgetline(); |
|---|
| 19 | char *embl_origin(), *genbank_origin(), *macke_origin(); |
|---|
| 20 | void init_seq_data(), gcg_seq_out(), gcg_doc_out(); |
|---|
| 21 | void embl_key_word(), genbank_key_word(); |
|---|
| 22 | void gcg_output_filename(); |
|---|
| 23 | int seqdata, Lenstr(), Cmpstr(), macke_abbrev(); |
|---|
| 24 | |
|---|
| 25 | if(intype==MACKE) { |
|---|
| 26 | if((ifp1=fopen(inf, "r"))==NULL||(ifp2=fopen(inf, "r"))==NULL |
|---|
| 27 | ||(ifp3=fopen(inf, "r"))==NULL) { |
|---|
| 28 | sprintf(temp, "Cannot open input file %s\n", inf); |
|---|
| 29 | error(38, temp); |
|---|
| 30 | } |
|---|
| 31 | } else if((ifp1=fopen(inf, "r"))==NULL) { |
|---|
| 32 | sprintf(temp, "CANNOT open input file %s, exit.\n", inf); |
|---|
| 33 | error(37, temp); |
|---|
| 34 | } |
|---|
| 35 | if(intype==MACKE) { |
|---|
| 36 | /* skip to #=; where seq. first appears */ |
|---|
| 37 | for(eof1=Fgetline(line1, LINENUM, ifp1); |
|---|
| 38 | eof1!=NULL&&(line1[0]!='#'||line1[1]!='='); |
|---|
| 39 | eof1=Fgetline(line1, LINENUM, ifp1)) ; |
|---|
| 40 | /* skip to #:; where the seq. information is */ |
|---|
| 41 | for(eof2=Fgetline(line2, LINENUM, ifp2); |
|---|
| 42 | eof2!=NULL&&(line2[0]!='#'||line2[1]!=':'); |
|---|
| 43 | eof2=Fgetline(line2, LINENUM, ifp2)) ; |
|---|
| 44 | /* skip to where seq. data starts */ |
|---|
| 45 | for(eof3=Fgetline(line3,LINENUM, ifp3); |
|---|
| 46 | eof3!=NULL&&line3[0]=='#'; |
|---|
| 47 | eof3=Fgetline(line3, LINENUM, ifp3)) ; |
|---|
| 48 | /* for each seq. print out one gcg file. */ |
|---|
| 49 | for(; eof1!=NULL&&(line1[0]=='#'&&line1[1]=='='); |
|---|
| 50 | eof1=Fgetline(line1, LINENUM, ifp1)) { |
|---|
| 51 | macke_abbrev(line1, key, 2); |
|---|
| 52 | Cpystr(temp, key); |
|---|
| 53 | gcg_output_filename(temp, outf); |
|---|
| 54 | if((ofp=fopen(outf, "w"))==NULL) { |
|---|
| 55 | sprintf(temp, "CANNOT open file %s, exit.\n", outf); |
|---|
| 56 | error(39, temp); |
|---|
| 57 | } |
|---|
| 58 | for(macke_abbrev(line2, name, 2); |
|---|
| 59 | eof2!=NULL&&line2[0]=='#'&&line2[1]==':'&&Cmpstr(name, key)==EQ; |
|---|
| 60 | eof2=Fgetline(line2, LINENUM, ifp2), macke_abbrev(line2, name, 2)) |
|---|
| 61 | gcg_doc_out(line2, ofp); |
|---|
| 62 | eof3 = macke_origin(key, line3, ifp3); |
|---|
| 63 | gcg_seq_out(ofp, key); |
|---|
| 64 | fclose(ofp); |
|---|
| 65 | init_seq_data(); |
|---|
| 66 | } |
|---|
| 67 | } else { |
|---|
| 68 | seqdata=0; /* flag of doc or data */ |
|---|
| 69 | eof = Fgetline(line, LINENUM, ifp1); |
|---|
| 70 | while(eof!=NULL) { |
|---|
| 71 | if(intype==GENBANK) { |
|---|
| 72 | genbank_key_word(line, 0, temp, TOKENNUM); |
|---|
| 73 | if((Cmpstr(temp, "LOCUS"))==EQ) { |
|---|
| 74 | genbank_key_word(line+12, 0, key, TOKENNUM); |
|---|
| 75 | gcg_output_filename(key, outf); |
|---|
| 76 | if((ofp=fopen(outf, "w"))==NULL) { |
|---|
| 77 | sprintf(temp, "CANNOT open file %s, exit.\n", outf); |
|---|
| 78 | error(40, temp); |
|---|
| 79 | } |
|---|
| 80 | } else if(Cmpstr(temp, "ORIGIN")==EQ) { |
|---|
| 81 | gcg_doc_out(line, ofp); |
|---|
| 82 | seqdata = 1; |
|---|
| 83 | eof = genbank_origin(line, ifp1); |
|---|
| 84 | } |
|---|
| 85 | } else { /* EMBL or SwissProt */ |
|---|
| 86 | if(Lenstr(line)>2&&line[0]=='I'&&line[1]=='D') |
|---|
| 87 | { |
|---|
| 88 | embl_key_word(line, 5, key, TOKENNUM); |
|---|
| 89 | gcg_output_filename(key, outf); |
|---|
| 90 | if((ofp=fopen(outf, "w"))==NULL) { |
|---|
| 91 | sprintf(temp, "CANNOT open file %s, exit.\n", outf); |
|---|
| 92 | error(41, temp); |
|---|
| 93 | } |
|---|
| 94 | } else if(Lenstr(line)>1&&line[0]=='S' |
|---|
| 95 | &&line[1]=='Q') { |
|---|
| 96 | gcg_doc_out(line, ofp); |
|---|
| 97 | seqdata = 1; |
|---|
| 98 | eof = embl_origin(line, ifp1); |
|---|
| 99 | } |
|---|
| 100 | } |
|---|
| 101 | if(seqdata) { |
|---|
| 102 | gcg_seq_out(ofp, key); |
|---|
| 103 | init_seq_data(); |
|---|
| 104 | seqdata=0; |
|---|
| 105 | fclose(ofp); |
|---|
| 106 | } else { |
|---|
| 107 | gcg_doc_out(line, ofp); |
|---|
| 108 | eof = Fgetline(line, LINENUM, ifp1); |
|---|
| 109 | } |
|---|
| 110 | } |
|---|
| 111 | } |
|---|
| 112 | } |
|---|
| 113 | /* ---------------------------------------------------------------- */ |
|---|
| 114 | /* Function gcg_seq_out(). |
|---|
| 115 | /* Output sequence data in gcg format. |
|---|
| 116 | */ |
|---|
| 117 | void |
|---|
| 118 | gcg_seq_out(ofp, key) |
|---|
| 119 | FILE *ofp; |
|---|
| 120 | char *key; |
|---|
| 121 | { |
|---|
| 122 | int checksum(), gcg_seq_length(), indi, indj, indk; |
|---|
| 123 | char *today_date(), *gcg_date(); |
|---|
| 124 | void gcg_out_origin(); |
|---|
| 125 | |
|---|
| 126 | fprintf(ofp, "\n%s Length: %d %s Type: N Check: %d ..\n\n", key, |
|---|
| 127 | gcg_seq_length(), gcg_date(today_date()), |
|---|
| 128 | checksum(data.sequence, data.seq_length)); |
|---|
| 129 | gcg_out_origin(ofp); |
|---|
| 130 | } |
|---|
| 131 | /* -------------------------------------------------------------------- */ |
|---|
| 132 | /* Function gcg_doc_out(). |
|---|
| 133 | /* Output non-sequence data(document) of gcg format. |
|---|
| 134 | */ |
|---|
| 135 | void |
|---|
| 136 | gcg_doc_out(line, ofp) |
|---|
| 137 | char *line; |
|---|
| 138 | FILE *ofp; |
|---|
| 139 | { |
|---|
| 140 | int indi, len, Lenstr(); |
|---|
| 141 | int previous_is_dot; |
|---|
| 142 | |
|---|
| 143 | for(indi=0, len=Lenstr(line), previous_is_dot=0; indi<len; indi++) { |
|---|
| 144 | if(previous_is_dot) { |
|---|
| 145 | if(line[indi]=='.') fprintf(ofp, " "); |
|---|
| 146 | else previous_is_dot=0; |
|---|
| 147 | } |
|---|
| 148 | fprintf(ofp, "%c", line[indi]); |
|---|
| 149 | if(line[indi]=='.') previous_is_dot=1; |
|---|
| 150 | } |
|---|
| 151 | } |
|---|
| 152 | /* ----------------------------------------------------------------- */ |
|---|
| 153 | /* Function checksum(). |
|---|
| 154 | /* Calculate checksum for GCG format. |
|---|
| 155 | */ |
|---|
| 156 | int |
|---|
| 157 | checksum(string, numofstr) |
|---|
| 158 | char *string; |
|---|
| 159 | int numofstr; |
|---|
| 160 | { |
|---|
| 161 | int cksum=0, indi, count=0, charnum; |
|---|
| 162 | |
|---|
| 163 | for(indi=0; indi<numofstr; indi++) { |
|---|
| 164 | if(string[indi]=='.'||string[indi]=='-'||string[indi]=='~') |
|---|
| 165 | continue; |
|---|
| 166 | count++; |
|---|
| 167 | if(string[indi]>='a'&&string[indi]<='z') |
|---|
| 168 | charnum = string[indi]-'a'+'A'; |
|---|
| 169 | else charnum = string[indi]; |
|---|
| 170 | cksum = ((cksum+count*charnum) % 10000); |
|---|
| 171 | if(count==57) count=0; |
|---|
| 172 | } |
|---|
| 173 | return(cksum); |
|---|
| 174 | } |
|---|
| 175 | /* -------------------------------------------------------------------- */ |
|---|
| 176 | /* Fcuntion gcg_out_origin(). |
|---|
| 177 | /* Output sequence data in gcg format. |
|---|
| 178 | */ |
|---|
| 179 | void |
|---|
| 180 | gcg_out_origin(fp) |
|---|
| 181 | FILE *fp; |
|---|
| 182 | { |
|---|
| 183 | |
|---|
| 184 | int indi, indj, indk; |
|---|
| 185 | |
|---|
| 186 | for(indi=0, indj=0, indk=1; indi<data.seq_length; indi++) { |
|---|
| 187 | if(data.sequence[indi]=='.'||data.sequence[indi]=='~' |
|---|
| 188 | ||data.sequence[indi]=='-') continue; |
|---|
| 189 | if((indk % 50)==1) fprintf(fp, "%8d ", indk); |
|---|
| 190 | fprintf(fp, "%c", data.sequence[indi]); |
|---|
| 191 | indj++; |
|---|
| 192 | if(indj==10) { fprintf(fp, " "); indj=0; } |
|---|
| 193 | if((indk % 50)==0) fprintf(fp, "\n\n"); |
|---|
| 194 | indk++; |
|---|
| 195 | } |
|---|
| 196 | if((indk % 50)!=1) fprintf(fp, " \n"); |
|---|
| 197 | } |
|---|
| 198 | /* -------------------------------------------------------------- */ |
|---|
| 199 | /* Function gcg_output_filename(). |
|---|
| 200 | /* Get gcg output filename, convert all '.' to '_' and |
|---|
| 201 | /* append ".RDP" as suffix. |
|---|
| 202 | */ |
|---|
| 203 | void |
|---|
| 204 | gcg_output_filename(prefix, name) |
|---|
| 205 | char *prefix, *name; |
|---|
| 206 | { |
|---|
| 207 | int indi, len, Lenstr(); |
|---|
| 208 | |
|---|
| 209 | for(indi=0, len=Lenstr(prefix); indi<len; indi++) |
|---|
| 210 | if(prefix[indi]=='.') prefix[indi]='_'; |
|---|
| 211 | sprintf(name, "%s.RDP", prefix); |
|---|
| 212 | } |
|---|
| 213 | /* ------------------------------------------------------------------ */ |
|---|
| 214 | /* Function gcg_seq_length(). |
|---|
| 215 | /* Calculate sequence length without gap. |
|---|
| 216 | */ |
|---|
| 217 | int |
|---|
| 218 | gcg_seq_length() { |
|---|
| 219 | |
|---|
| 220 | int indi, len; |
|---|
| 221 | |
|---|
| 222 | for(indi=0, len=data.seq_length; indi<data.seq_length; indi++) |
|---|
| 223 | if(data.sequence[indi]=='.'||data.sequence[indi]=='-' |
|---|
| 224 | ||data.sequence[indi]=='~') len--; |
|---|
| 225 | |
|---|
| 226 | return(len); |
|---|
| 227 | } |
|---|
