Mercurial > repos > nick > duplex
annotate seqtools.py @ 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 import os |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
2 import ctypes |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
3 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
4 script_dir = os.path.dirname(os.path.realpath(__file__)) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
5 seqtools = ctypes.cdll.LoadLibrary(os.path.join(script_dir, 'libseqtools.so')) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
6 seqtools.get_revcomp.restype = ctypes.c_char_p |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
7 seqtools.transfer_gaps.restype = ctypes.c_char_p |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
8 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
9 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
10 def get_revcomp(seq): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
11 return seqtools.get_revcomp(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
12 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
13 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
14 def get_diffs_frac_simple(consensus, family): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
15 c_consensus = ctypes.c_char_p(consensus) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
16 c_family = (ctypes.c_char_p * len(family))() |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
17 for i, seq in enumerate(family): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
18 c_family[i] = ctypes.c_char_p(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
19 seqtools.get_diffs_frac_simple.restype = ctypes.POINTER(ctypes.c_double * len(c_family)) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
20 diffs = seqtools.get_diffs_frac_simple(c_consensus, c_family, len(c_family)) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
21 return diffs.contents |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
22 |
|
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 def get_diffs_frac_binned(consensus, family, bins): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
25 seq_len = None |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
26 c_consensus = ctypes.c_char_p(consensus) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
27 c_family = (ctypes.c_char_p * len(family))() |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
28 for i, seq in enumerate(family): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
29 if seq_len: |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
30 if seq_len != len(seq): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
31 return None |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
32 else: |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
33 seq_len = len(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
34 c_family[i] = ctypes.c_char_p(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
35 double_array_pointer = ctypes.POINTER(ctypes.c_double * bins) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
36 seqtools.get_diffs_frac_binned.restype = ctypes.POINTER(double_array_pointer * len(c_family)) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
37 diffs_binned_c = seqtools.get_diffs_frac_binned(c_consensus, c_family, len(c_family), seq_len, bins) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
38 diffs_binned = [] |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
39 for diffs_c in diffs_binned_c.contents: |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
40 diffs_binned.append(diffs_c.contents) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
41 return diffs_binned |
|
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 def transfer_gaps(aligned, seq, gap_char_in='-', gap_char_out='-'): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
45 gap_char_in_c = ctypes.c_char(gap_char_in) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
46 gap_char_out_c = ctypes.c_char(gap_char_out) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
47 return seqtools.transfer_gaps(aligned, seq, gap_char_in_c, gap_char_out_c) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
48 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
49 |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
50 def transfer_gaps_multi(seqs, aligned, gap_char_in='-', gap_char_out='-'): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
51 gap_char_in_c = ctypes.c_char(gap_char_in) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
52 gap_char_out_c = ctypes.c_char(gap_char_out) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
53 n_seqs = len(seqs) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
54 assert n_seqs == len(aligned), 'Error: Unequal number of gapped and ungapped sequences.' |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
55 seqs_c = (ctypes.c_char_p * n_seqs)() |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
56 for i, seq in enumerate(seqs): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
57 seqs_c[i] = ctypes.c_char_p(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
58 aligned_c = (ctypes.c_char_p * n_seqs)() |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
59 for i, seq in enumerate(aligned): |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
60 aligned_c[i] = ctypes.c_char_p(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
61 seqtools.transfer_gaps_multi.restype = ctypes.POINTER(ctypes.c_char_p * n_seqs) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
62 output_c = seqtools.transfer_gaps_multi(n_seqs, aligned_c, seqs_c, gap_char_in_c, gap_char_out_c) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
63 output = [] |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
64 for seq in output_c.contents: |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
65 output.append(seq) |
|
e4d75f9efb90
planemo upload commit b'4303231da9e48b2719b4429a29b72421d24310f4\n'-dirty
nick
parents:
diff
changeset
|
66 return output |
