annotate seqtools.c @ 18:e4d75f9efb90 draft

planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
author nick
date Thu, 02 Feb 2017 18:44:31 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
18
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
1 #include <stdlib.h>
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
2 #include <string.h>
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
3 #include <ctype.h>
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
4 #include <math.h>
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
5
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
6 // ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
7 #define TRANS "TVGHEFCDIJMLKNOPQYWAABSXRZ[\\]^_`tvghefcdijmlknopqywaabsxrz"
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
8 #define TRANS_OFFSET 65
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
9 #define TRANS_LEN 57
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
10
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
11 char* get_revcomp(char *input);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
12 char get_char_comp(char c);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
13 int *get_diffs_simple(char *cons, char *seqs[], int n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
14 double *get_diffs_frac_simple(char *cons, char *seqs[], int n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
15 double **get_diffs_frac_binned(char *cons, char *seqs[], int n_seqs, int seq_len, int bins);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
16 char *transfer_gaps(char *gapped_seq, char *inseq, char gap_char1, char gap_char2);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
17 char **transfer_gaps_multi(int n_seqs, char *gapped_seqs[], char *inseqs[], char gap_char1, char gap_char2);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
18
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
19
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
20 // Return the reverse complement of a sequence.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
21 // Makes a new copy of the string, so the original is not modified.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
22 char* get_revcomp(char *input) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
23 int length = strlen(input);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
24 char *output = malloc(sizeof(char) * length + 1);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
25 int i, j;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
26 for (i = 0, j = length - 1; i < length && j >= 0; i++, j--) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
27 output[j] = get_char_comp(input[i]);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
28 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
29 output[length] = '\0';
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
30 return output;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
31 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
32
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
33
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
34 // Return the complement of a base.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
35 // Uses a simple lookup table: a string with the complements of all possible sequence characters.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
36 char get_char_comp(char c) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
37 int i = c - TRANS_OFFSET;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
38 if (i < 0 || i > TRANS_LEN) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
39 return c;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
40 } else {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
41 return TRANS[i];
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
42 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
43 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
44
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
45
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
46 /* Take an existing alignment and consensus and compute the number of differences between each
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
47 * sequence and the consensus.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
48 * Known bugs:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
49 * 1. Counts no differences in the following sequences:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
50 * consensus: GA---CA
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
51 * seq 1: GA----A
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
52 * seq 2: GA--ACA
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
53 * 2. If a sequence starts with a gap, each base in the gap will be counted as a diff.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
54 */
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
55 int *get_diffs_simple(char *cons, char *seqs[], int n_seqs) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
56 int *diffs = malloc(sizeof(int) * n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
57 int i = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
58 // Uppercase the consensus.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
59 while (cons[i] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
60 cons[i] = toupper(cons[i]);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
61 i++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
62 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
63 // Loop through the sequences in the alignment.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
64 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
65 int in_gap;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
66 diffs[i] = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
67 int j = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
68 // Compare each base of the sequence to the consensus.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
69 while (seqs[i][j] != 0 && cons[j] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
70 if (cons[j] != '-' && seqs[i][j] != '-') {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
71 in_gap = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
72 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
73 if (toupper(seqs[i][j]) != cons[j]) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
74 if (!in_gap) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
75 diffs[i]++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
76 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
77 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
78 if (cons[j] == '-' || seqs[i][j] == '-') {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
79 in_gap = 1;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
80 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
81 j++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
82 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
83 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
84 return diffs;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
85 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
86
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
87
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
88 // Convert the output of get_diffs_simple() from raw diff counts to fractions of the total sequence
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
89 // lengths.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
90 //TODO: Don't count gaps in sequence length.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
91 double *get_diffs_frac_simple(char *cons, char *seqs[], int n_seqs) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
92 int *diffs = get_diffs_simple(cons, seqs, n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
93 double *fracs = malloc(sizeof(double) * n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
94 int i;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
95 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
96 int j = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
97 while (seqs[i][j] != 0 && cons[j] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
98 j++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
99 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
100 fracs[i] = (double)diffs[i]/j;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
101 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
102 return fracs;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
103 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
104
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
105
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
106 /* Take an existing alignment and consensus and compute the number of differences between each
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
107 * sequence and the consensus. Break each sequence into bins and tally the differences in each bin.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
108 * Known bugs:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
109 * 1. counts no differences in the following sequences:
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
110 * consensus: GA---CA
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
111 * seq 1: GA----A
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
112 * seq 2: GA--ACA
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
113 * 2. If a bin starts with a gap, each base in the gap will be counted as a diff.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
114 */
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
115 int **get_diffs_binned(char *cons, char *seqs[], int n_seqs, int seq_len, int bins) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
116 int bin_size = (int)round((float)seq_len/bins);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
117 // Initialize the diffs 2d array.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
118 int **diffs = malloc(sizeof(int*) * n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
119 int i, j;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
120 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
121 diffs[i] = malloc(bins * sizeof(int));
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
122 for (j = 0; j < bins; j++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
123 diffs[i][j] = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
124 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
125 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
126 // Uppercase the consensus.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
127 while (cons[i] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
128 cons[i] = toupper(cons[i]);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
129 i++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
130 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
131 int bin, in_gap;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
132 // Loop through the sequences in the alignment.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
133 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
134 j = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
135 // Compare each base of the sequence to the consensus.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
136 while (seqs[i][j] != 0 && cons[j] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
137 bin = j/bin_size;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
138 if (bin >= bins) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
139 break;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
140 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
141 if (cons[j] != '-' && seqs[i][j] != '-') {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
142 in_gap = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
143 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
144 if (toupper(seqs[i][j]) != cons[j]) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
145 if (!in_gap) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
146 diffs[i][bin]++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
147 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
148 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
149 if (cons[j] == '-' || seqs[i][j] == '-') {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
150 in_gap = 1;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
151 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
152 j++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
153 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
154 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
155 return diffs;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
156 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
157
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
158
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
159 // Convert the output of get_diffs_binned() from raw diff counts to fractions of the total bin
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
160 // lengths.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
161 //TODO: Don't count gaps in bin length.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
162 double **get_diffs_frac_binned(char *cons, char *seqs[], int n_seqs, int seq_len, int bins) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
163 int bin_size = (int)round((float)seq_len/bins);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
164 int **diffs = get_diffs_binned(cons, seqs, n_seqs, seq_len, bins);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
165 double **fracs = malloc(sizeof(double*) * n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
166 int i;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
167 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
168 fracs[i] = malloc(sizeof(double) * bins);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
169 // Create and init array of lengths of the bins.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
170 int bin_lengths[bins];
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
171 int bin;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
172 for (bin = 0; bin < bins; bin++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
173 bin_lengths[bin] = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
174 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
175 // Tally size of each bin.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
176 int j = 0;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
177 while (seqs[i][j] != 0 && cons[j] != 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
178 int bin = j/bin_size;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
179 if (bin >= bins) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
180 break;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
181 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
182 bin_lengths[bin]++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
183 j++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
184 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
185 // For each bin, calculate the diff fraction = diffs / bin_length.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
186 for (bin = 0; bin < bins; bin++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
187 fracs[i][bin] = (double)diffs[i][bin]/bin_lengths[bin];
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
188 // printf("bin %d: %d / %d = %f\t", bin, diffs[i][bin], bin_lengths[bin], fracs[i][bin]);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
189 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
190 // printf("\n");
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
191 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
192 return fracs;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
193 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
194
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
195
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
196 // Take an input sequence and insert gaps according to another, already-aligned sequence with gaps.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
197 // Input strings must be null-terminated. "gap_char1" is the character used for gaps in
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
198 // "gapped_seq", and "gap_char2" is the gap character in "inseq".
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
199 // N.B.: The ungapped length of "gapped_seq" must be equal to the length of "inseq".
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
200 char *transfer_gaps(char *gapped_seq, char *inseq, char gap_char1, char gap_char2) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
201 if (gap_char1 == 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
202 gap_char1 = '-';
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
203 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
204 if (gap_char2 == 0) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
205 gap_char2 = '-';
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
206 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
207 int gapped_len = strlen(gapped_seq);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
208 char *outseq = malloc(sizeof(char) * gapped_len + 1);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
209
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
210 // Transfer characters from inseq to outseq, except when gapped_seq has a gap at that spot
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
211 // (insert a gap there instead).
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
212 int g, o, i;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
213 for (g = 0, o = 0, i = 0; g < gapped_len; g++, o++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
214 if (gapped_seq[g] == gap_char1) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
215 outseq[o] = gap_char2;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
216 } else {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
217 outseq[o] = inseq[i];
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
218 i++;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
219 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
220 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
221 outseq[gapped_len] = '\0';
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
222
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
223 return outseq;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
224 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
225
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
226
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
227 // Wrapper for transfer_gaps() when operating on a set of sequences at once.
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
228 char **transfer_gaps_multi(int n_seqs, char *gapped_seqs[], char *inseqs[], char gap_char1,
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
229 char gap_char2) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
230 char **outseqs = malloc(sizeof(char *) * n_seqs);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
231 int i;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
232 for (i = 0; i < n_seqs; i++) {
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
233 outseqs[i] = transfer_gaps(gapped_seqs[i], inseqs[i], gap_char1, gap_char2);
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
234 }
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
235 return outseqs;
e4d75f9efb90 planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff changeset
236 }