1 | """ |
---|
2 | Qualityscore class |
---|
3 | """ |
---|
4 | |
---|
5 | import data |
---|
6 | import logging |
---|
7 | from galaxy.datatypes.sniff import * |
---|
8 | from galaxy import util |
---|
9 | |
---|
10 | log = logging.getLogger(__name__) |
---|
11 | |
---|
12 | class QualityScore ( data.Text ): |
---|
13 | """ |
---|
14 | until we know more about quality score formats |
---|
15 | """ |
---|
16 | file_ext = "qual" |
---|
17 | |
---|
18 | class QualityScoreSOLiD ( QualityScore ): |
---|
19 | """ |
---|
20 | until we know more about quality score formats |
---|
21 | """ |
---|
22 | file_ext = "qualsolid" |
---|
23 | |
---|
24 | def sniff( self, filename ): |
---|
25 | """ |
---|
26 | >>> fname = get_test_fname( 'sequence.fasta' ) |
---|
27 | >>> QualityScoreSOLiD().sniff( fname ) |
---|
28 | False |
---|
29 | >>> fname = get_test_fname( 'sequence.qualsolid' ) |
---|
30 | >>> QualityScoreSOLiD().sniff( fname ) |
---|
31 | True |
---|
32 | """ |
---|
33 | try: |
---|
34 | fh = open( filename ) |
---|
35 | readlen = None |
---|
36 | goodblock = 0 |
---|
37 | while True: |
---|
38 | line = fh.readline() |
---|
39 | if not line: |
---|
40 | if goodblock > 0: |
---|
41 | return True |
---|
42 | else: |
---|
43 | break #EOF |
---|
44 | line = line.strip() |
---|
45 | if line and not line.startswith( '#' ): #first non-empty non-comment line |
---|
46 | if line.startswith( '>' ): |
---|
47 | line = fh.readline().strip() |
---|
48 | if line == '' or line.startswith( '>' ): |
---|
49 | break |
---|
50 | try: |
---|
51 | [ int( x ) for x in line.split() ] |
---|
52 | if not(readlen): |
---|
53 | readlen = len(line.split()) |
---|
54 | assert len(line.split()) == readlen #SOLiD reads should be of the same length |
---|
55 | except: |
---|
56 | break |
---|
57 | goodblock += 1 |
---|
58 | if goodblock > 10: |
---|
59 | return True |
---|
60 | else: |
---|
61 | break #we found a non-empty line, but it's not a header |
---|
62 | fh.close() |
---|
63 | except: |
---|
64 | pass |
---|
65 | return False |
---|
66 | |
---|
67 | def set_meta( self, dataset, **kwd ): |
---|
68 | if self.max_optional_metadata_filesize >= 0 and dataset.get_size() > self.max_optional_metadata_filesize: |
---|
69 | dataset.metadata.data_lines = None |
---|
70 | return |
---|
71 | return QualityScore.set_meta( self, dataset, **kwd ) |
---|
72 | |
---|
73 | |
---|
74 | |
---|
75 | class QualityScore454 ( QualityScore ): |
---|
76 | """ |
---|
77 | until we know more about quality score formats |
---|
78 | """ |
---|
79 | file_ext = "qual454" |
---|
80 | |
---|
81 | def sniff( self, filename ): |
---|
82 | """ |
---|
83 | >>> fname = get_test_fname( 'sequence.fasta' ) |
---|
84 | >>> QualityScore454().sniff( fname ) |
---|
85 | False |
---|
86 | >>> fname = get_test_fname( 'sequence.qual454' ) |
---|
87 | >>> QualityScore454().sniff( fname ) |
---|
88 | True |
---|
89 | """ |
---|
90 | try: |
---|
91 | fh = open( filename ) |
---|
92 | while True: |
---|
93 | line = fh.readline() |
---|
94 | if not line: |
---|
95 | break #EOF |
---|
96 | line = line.strip() |
---|
97 | if line and not line.startswith( '#' ): #first non-empty non-comment line |
---|
98 | if line.startswith( '>' ): |
---|
99 | line = fh.readline().strip() |
---|
100 | if line == '' or line.startswith( '>' ): |
---|
101 | break |
---|
102 | try: |
---|
103 | [ int( x ) for x in line.split() ] |
---|
104 | except: |
---|
105 | break |
---|
106 | return True |
---|
107 | else: |
---|
108 | break #we found a non-empty line, but it's not a header |
---|
109 | fh.close() |
---|
110 | except: |
---|
111 | pass |
---|
112 | return False |
---|
113 | |
---|
114 | class QualityScoreSolexa ( QualityScore ): |
---|
115 | """ |
---|
116 | until we know more about quality score formats |
---|
117 | """ |
---|
118 | file_ext = "qualsolexa" |
---|
119 | |
---|
120 | class QualityScoreIllumina ( QualityScore ): |
---|
121 | """ |
---|
122 | until we know more about quality score formats |
---|
123 | """ |
---|
124 | file_ext = "qualillumina" |
---|