Mercurial > repos > nick > duplex
annotate align.c @ 18:e4d75f9efb90 draft
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
author | nick |
---|---|
date | Thu, 02 Feb 2017 18:44:31 -0500 |
parents | |
children |
rev | line source |
---|---|
18
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
1 #include <stdio.h> |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
2 #include <stdlib.h> |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
3 #include <string.h> |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
4 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
5 #define NAIVE_TEST_WINDOW 6 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
6 #define NAIVE_TEST_THRES 0.80 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
7 #define NAIVE_TEST_MIN 2 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
8 #define NAIVE_WINDOW 10 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
9 #define NAIVE_THRES 0.80 |
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 typedef struct Gap { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
12 int seq; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
13 int coord; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
14 int length; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
15 struct Gap *next; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
16 } Gap; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
17 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
18 typedef struct Gaps { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
19 int length; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
20 struct Gap *root; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
21 struct Gap *tip; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
22 } Gaps; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
23 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
24 int _test_match(char *seq1, int start1, char *seq2, int start2); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
25 void add_gap(Gaps *gaps, int seq, int coord, int length); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
26 Gaps *make_gaps(); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
27 char *insert_gaps(Gaps *gaps, char *seq, int seq_num); |
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 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
30 // A naive algorithm for aligning two sequences which are expected to be very similar to each other |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
31 // and already nearly aligned. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
32 void naive2(char *seq1, char *seq2) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
33 Gaps *gaps = make_gaps(); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
34 int i = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
35 int j = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
36 int matches = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
37 while (seq1[i] != 0 && seq2[j] != 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
38 // Match? |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
39 printf("%c %c | i %d j %d\n", seq1[i], seq2[j], i, j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
40 if (seq1[i] == seq2[j]) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
41 matches++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
42 i++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
43 j++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
44 continue; |
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 printf("mismatch!\n"); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
47 // Mismatch. Start adding gaps until the mismatches go away. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
48 int new_i = i; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
49 int new_j = j; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
50 int gap_seq = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
51 int success; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
52 while (1) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
53 if (seq1[new_i] == 0 && seq2[new_j] == 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
54 break; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
55 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
56 success = _test_match(seq1, new_i, seq2, j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
57 if (success) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
58 gap_seq = 2; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
59 break; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
60 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
61 if (seq1[new_i] != 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
62 new_i++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
63 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
64 success = _test_match(seq1, i, seq2, new_j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
65 if (success) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
66 gap_seq = 1; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
67 break; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
68 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
69 if (seq2[new_j] != 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
70 new_j++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
71 } |
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 // Which sequence are we putting the gap in? |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
74 if (gap_seq == 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
75 printf("No good gap found. new_i: %d, new_j: %d\n", new_i, new_j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
76 // No good gap found. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
77 } else if (i == new_i && j == new_j) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
78 printf("No gap required.\n"); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
79 } else if (gap_seq == 1) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
80 printf("%dbp gap in seq1 at base %d.\n", new_j-j, j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
81 add_gap(gaps, 1, j, new_j-j); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
82 j = new_j; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
83 } else if (gap_seq == 2) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
84 printf("%dbp gap in seq2 at base %d.\n", new_i-i, i); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
85 add_gap(gaps, 2, i, new_i-i); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
86 i = new_i; |
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 i++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
89 j++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
90 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
91 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
92 char *new_seq1 = insert_gaps(gaps, seq1, 1); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
93 char *new_seq2 = insert_gaps(gaps, seq2, 2); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
94 printf("alignment:\n%s\n%s\n", new_seq1, new_seq2); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
95 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
96 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
97 // Check if the few bases starting at start1 and start2 in seq1 and seq2, respectively, align with |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
98 // few mismatches. The number of bases checked is NAIVE_TEST_WINDOW, and they must have a match |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
99 // percentage greater than NAIVE_TEST_THRES. Also, the amount of sequence left to compare must be |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
100 // more than NAIVE_TEST_MIN. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
101 int _test_match(char *seq1, int start1, char *seq2, int start2) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
102 int matches = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
103 int total = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
104 char base1, base2; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
105 int i; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
106 for (i = 0; i < NAIVE_TEST_WINDOW-1; i++) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
107 base1 = seq1[start1+i]; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
108 base2 = seq2[start2+i]; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
109 if (base1 == 0 || base2 == 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
110 break; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
111 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
112 if (base1 == base2) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
113 matches++; |
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 total++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
116 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
117 return total > NAIVE_TEST_MIN && (double)matches/total > NAIVE_TEST_THRES; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
118 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
119 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
120 Gaps *make_gaps() { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
121 Gaps *gaps = malloc(sizeof(Gaps)); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
122 gaps->root = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
123 gaps->tip = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
124 gaps->length = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
125 return gaps; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
126 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
127 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
128 void add_gap(Gaps *gaps, int seq, int coord, int length) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
129 Gap *gap = malloc(sizeof(Gap)); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
130 gap->next = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
131 gap->seq = seq; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
132 gap->coord = coord; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
133 gap->length = length; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
134 if (gaps->root == 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
135 gaps->root = gap; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
136 } else { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
137 gaps->tip->next = gap; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
138 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
139 gaps->tip = gap; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
140 gaps->length++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
141 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
142 |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
143 // Take gap information from the aligner and put them into the sequence string as "-" characters. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
144 char *insert_gaps(Gaps *gaps, char *seq, int seq_num) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
145 if (gaps->root == 0) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
146 return seq; |
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 // How long should the new sequence be? |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
150 int extra_len = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
151 Gap *gap = gaps->root; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
152 while (gap) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
153 if (gap->seq == seq_num) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
154 extra_len += gap->length; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
155 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
156 gap = gap->next; |
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 //TODO: Handle a situation with no gaps. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
160 int new_len = extra_len + strlen(seq) + 1; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
161 char *new_seq = malloc(sizeof(char) * new_len); |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
162 int i = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
163 int j = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
164 gap = gaps->root; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
165 while (gap) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
166 // Check that it's a gap in our sequence. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
167 if (gap->seq != seq_num) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
168 gap = gap->next; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
169 continue; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
170 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
171 // Copy verbatim all the sequence until the gap. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
172 while (i <= gap->coord) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
173 new_seq[j] = seq[i]; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
174 i++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
175 j++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
176 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
177 // Add -'s the whole length of the gap. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
178 while (j < gap->coord + gap->length + 1) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
179 new_seq[j] = '-'; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
180 j++; |
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 gap = gap->next; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
183 } |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
184 // Fill in the end sequence. |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
185 while (seq[i]) { |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
186 new_seq[j] = seq[i]; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
187 i++; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
188 j++; |
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 new_seq[new_len-1] = 0; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
191 return new_seq; |
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
192 } |