| 1 | /* |
|---|
| 2 | * Copyright 1991 Steven Smith at the Harvard Genome Lab. |
|---|
| 3 | * All rights reserved. |
|---|
| 4 | */ |
|---|
| 5 | #include "Flatio.c" |
|---|
| 6 | |
|---|
| 7 | int Comp(a,b) |
|---|
| 8 | char a,b; |
|---|
| 9 | { |
|---|
| 10 | static int CtoB[128]={ |
|---|
| 11 | 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0x00, |
|---|
| 12 | 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, |
|---|
| 13 | 0x01,0xe,0x02,0x0d,0,0,0x04,0x0b,0,0,0x0c,0,0x03,0x0f,0,0,0,0x05,0x06, |
|---|
| 14 | 0x08,0x08,0x07,0,0x09,0xa,0,0,0,0,0,0,0,0x01,0x0e,0x02,0x0d,0,0,0x04, |
|---|
| 15 | 0x0b,0,0,0x0c,0,0x03,0x0f,0,0,0,0x05,0x06,0x08,0x08,0x07,0,0x09,0x0a, |
|---|
| 16 | 0,0,0,0,0x00,0 |
|---|
| 17 | }; |
|---|
| 18 | |
|---|
| 19 | return ((CtoB[a]) & (CtoB[b])); |
|---|
| 20 | } |
|---|
| 21 | |
|---|
| 22 | int FindNext(target,offset,enzymes,numenzymes,match_len,color) |
|---|
| 23 | char *target,enzymes[][80]; |
|---|
| 24 | int offset,numenzymes,*match_len,*color; |
|---|
| 25 | { |
|---|
| 26 | int i,j,k,closest,len1,dif,flag = FALSE; |
|---|
| 27 | closest = strlen(target); |
|---|
| 28 | *match_len = 0; |
|---|
| 29 | for(k=0;k<numenzymes;k++) |
|---|
| 30 | { |
|---|
| 31 | dif = (strlen(target)) - (len1 = strlen(enzymes[k])) +1; |
|---|
| 32 | |
|---|
| 33 | if(len1>0) |
|---|
| 34 | for(flag = FALSE,j=offset;j<dif && flag == FALSE;j++) |
|---|
| 35 | { |
|---|
| 36 | flag = TRUE; |
|---|
| 37 | for(i=0;i<len1 && flag;i++) |
|---|
| 38 | { |
|---|
| 39 | flag = Comp(enzymes[k][i],target[i+j])? |
|---|
| 40 | TRUE:FALSE; |
|---|
| 41 | } |
|---|
| 42 | } |
|---|
| 43 | if(j-1<closest) |
|---|
| 44 | { |
|---|
| 45 | closest = j-1; |
|---|
| 46 | *color = k%6+1; |
|---|
| 47 | *match_len = strlen(enzymes[k]); |
|---|
| 48 | } |
|---|
| 49 | } |
|---|
| 50 | if(closest + *match_len < (int)strlen(target)) |
|---|
| 51 | return(closest); |
|---|
| 52 | else |
|---|
| 53 | return(-1); |
|---|
| 54 | } |
|---|
| 55 | |
|---|
| 56 | int main(ac,av) |
|---|
| 57 | int ac; |
|---|
| 58 | char **av; |
|---|
| 59 | { |
|---|
| 60 | struct data_format data[10000]; |
|---|
| 61 | FILE *file; |
|---|
| 62 | int i,j,k,color,numseqs,numenzymes,nextpos,len; |
|---|
| 63 | char enzymes[80][80],dummy[80]; |
|---|
| 64 | if(ac<3) |
|---|
| 65 | { |
|---|
| 66 | fprintf(stderr,"Usage: %s enzyme_file seq_file\n",av[0]); |
|---|
| 67 | return -1; |
|---|
| 68 | } |
|---|
| 69 | file = fopen(av[2],"r"); |
|---|
| 70 | if(file == NULL) |
|---|
| 71 | return -1; |
|---|
| 72 | |
|---|
| 73 | numseqs = ReadFlat(file,data,10000); |
|---|
| 74 | |
|---|
| 75 | file = fopen(av[1],"r"); |
|---|
| 76 | if(file == NULL) |
|---|
| 77 | return -1; |
|---|
| 78 | |
|---|
| 79 | for(numenzymes = 0; |
|---|
| 80 | fscanf(file,"%s %s",enzymes[numenzymes],dummy)>0; |
|---|
| 81 | numenzymes++); |
|---|
| 82 | |
|---|
| 83 | for(i=0;i<numseqs;i++) |
|---|
| 84 | { |
|---|
| 85 | /* |
|---|
| 86 | if(numseqs>1) |
|---|
| 87 | */ |
|---|
| 88 | printf("name:%s\n",data[i].name); |
|---|
| 89 | printf("length:%zu\n",strlen(data[i].nuc)); |
|---|
| 90 | if(numseqs>1) |
|---|
| 91 | printf("nodash:\n"); |
|---|
| 92 | printf("start:\n"); |
|---|
| 93 | for(j=0;j<data[i].length;) |
|---|
| 94 | { |
|---|
| 95 | for(;data[i].nuc[j] == '-' && j<data[i].length;) |
|---|
| 96 | { |
|---|
| 97 | printf("8\n"); |
|---|
| 98 | j++; |
|---|
| 99 | } |
|---|
| 100 | if((nextpos = FindNext(data[i].nuc,j,enzymes,numenzymes |
|---|
| 101 | ,&len,&color)) != -1) |
|---|
| 102 | { |
|---|
| 103 | for(k=j;k<nextpos;k++) |
|---|
| 104 | printf("8\n"); |
|---|
| 105 | for(k=j+nextpos;k<j+nextpos+len;k++) |
|---|
| 106 | printf("%d\n",color); |
|---|
| 107 | j=nextpos+len; |
|---|
| 108 | } |
|---|
| 109 | else |
|---|
| 110 | for(;j<data[i].length;j++) |
|---|
| 111 | printf("8\n"); |
|---|
| 112 | } |
|---|
| 113 | } |
|---|
| 114 | return 0; |
|---|
| 115 | } |
|---|
| 116 | |
|---|
| 117 | |
|---|