Mercurial > repos > galaxyp > mzsqlite_psm_align

annotate profmt.py @ 1:4f8cf8fbef57 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit b0c57cac4e558d974a16b14d4498cf8d4ba9e0c7
author galaxyp
date Thu, 19 Apr 2018 14:30:28 -0400
parents f2dc9805107a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
2 """
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
3 #
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
4 #------------------------------------------------------------------------------
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
5 # University of Minnesota
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
6 # Copyright 2018, Regents of the University of Minnesota
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
7 #------------------------------------------------------------------------------
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
8 # Author:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
9 #
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
10 # James E Johnson
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
11 #
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
12 #------------------------------------------------------------------------------
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
13 """
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
14
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
15 import sys,re
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
16 from operator import itemgetter, attrgetter
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
17 from twobitreader import TwoBitFile
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
18
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
19
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
20 PROBAM_TAGS = ['NH', 'XO', 'XL', 'XP', 'YP', 'XF', 'XI', 'XB', 'XR', 'YB', 'YA', 'XS', 'XQ', 'XC', 'XA', 'XM', 'XN', 'XT', 'XE', 'XG', 'XU']
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
21
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
22
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
23 PROBAM_TYTPES = {
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
24 'NH' : 'i', #number of genomic locations to which the peptide sequence maps
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
25 'XO' : 'Z', #uniqueness of the peptide mapping
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
26 'XL' : 'i', #number of peptides to which the spectrum maps
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
27 'XP' : 'Z', #peptide sequence
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
28 'YP' : 'Z', #Protein accession ID from the original search result
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
29 'XF' : 'Z', #Reading frame of the peptide (0, 1, 2)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
30 'XI' : 'f', #Peptide intensity
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
31 'XB' : 'Z', #massdiff; experimental mass; calculated mass massdiff can be calculated by experimental mass - calculated mass. If any number is unavailable, the value should be left blank (such as 0.01;;).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
32 'XR' : 'Z', #reference peptide sequence
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
33 'YB' : 'Z', #Preceding amino acids (2 AA, B stands for before).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
34 'YA' : 'Z', #Following amino acids (2 AA, A stands for after).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
35 'XS' : 'f', #PSM score
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
36 'XQ' : 'f', #PSM FDR (i.e. q-value or 1-PEP).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
37 'XC' : 'i', #peptide charge
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
38 'XA' : 'i', #Whether the peptide is annotated 0:yes; 1:parially unknown; 2:totally unknown;
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
39 'XM' : 'Z', #Modifications
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
40 'XN' : 'i', #Number of missed cleavages in the peptide (XP)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
41 'XT' : 'i', #Enzyme specificity
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
42 'XE' : 'i', #Enzyme used in the experiment
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
43 'XG' : 'A', #Peptide type
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
44 'XU' : 'Z', #URI
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
45 }
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
46
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
47
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
48 PROBAM_DEFAULTS = {
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
49 'NH' : -1, #number of genomic locations to which the peptide sequence maps
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
50 'XO' : '*', #uniqueness of the peptide mapping
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
51 'XL' : -1, #number of peptides to which the spectrum maps
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
52 'XP' : '*', #peptide sequence
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
53 'YP' : '*', #Protein accession ID from the original search result
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
54 'XF' : '*', #Reading frame of the peptide (0, 1, 2)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
55 'XI' : -1, #Peptide intensity
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
56 'XB' : '*', #massdiff; experimental mass; calculated mass massdiff can be calculated by experimental mass - calculated mass. If any number is unavailable, the value should be left blank (such as 0.01;;).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
57 'XR' : '*', #reference peptide sequence
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
58 'YB' : '*', #Preceding amino acids (2 AA, B stands for before).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
59 'YA' : '*', #Following amino acids (2 AA, A stands for after).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
60 'XS' : -1, #PSM score
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
61 'XQ' : -1, #PSM FDR (i.e. q-value or 1-PEP).
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
62 'XC' : -1, #peptide charge
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
63 'XA' : -1, #Whether the peptide is annotated 0:yes; 1:parially unknown; 2:totally unknown;
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
64 'XM' : '*', #Modifications
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
65 'XN' : -1, #Number of missed cleavages in the peptide (XP)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
66 'XT' : -1, #Enzyme specificity
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
67 'XE' : -1, #Enzyme used in the experiment
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
68 'XG' : '*', #Peptide type
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
69 'XU' : '*', #URI
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
70 }
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
71
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
72 def cmp_alphanumeric(s1,s2):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
73 if s1 == s2:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
74 return 0
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
75 a1 = re.findall("\d+|[a-zA-Z]+",s1)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
76 a2 = re.findall("\d+|[a-zA-Z]+",s2)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
77 for i in range(min(len(a1),len(a2))):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
78 if a1[i] == a2[i]:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
79 continue
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
80 if a1[i].isdigit() and a2[i].isdigit():
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
81 return int(a1[i]) - int(a2[i])
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
82 return 1 if a1[i] > a2[i] else -1
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
83 return len(a1) - len(a2)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
84
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
85
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
86 def sort_chrom_names(names):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
87 rnames = sorted(names,cmp=cmp_alphanumeric)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
88 if 'chrM' in rnames:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
89 rnames.remove('chrM')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
90 rnames.insert(0,'chrM')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
91 if 'MT' in rnames:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
92 rnames.remove('MT')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
93 rnames.append('MT')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
94 return rnames
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
95
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
96
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
97 def as_int_list(obj):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
98 if obj is None:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
99 return None
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
100 if isinstance(obj, list):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
101 return [int(x) for x in obj]
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
102 elif isinstance(obj, str):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
103 return [int(x) for x in obj.split(',')]
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
104 else: # python2 unicode?
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
105 return [int(x) for x in str(obj).split(',')]
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
106
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
107
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
108 class ProBEDEntry (object):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
109 def __init__(self, chrom, chromStart, chromEnd, name, score, strand,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
110 blockCount, blockSizes, blockStarts,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
111 protacc, peptide, uniqueness, genomeReference,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
112 psmScore='.', fdr='.', mods='.', charge='.',
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
113 expMassToCharge='.', calcMassToCharge='.',
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
114 psmRank='.', datasetID='.', uri='.'):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
115 self.chrom = chrom
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
116 self.chromStart = int(chromStart)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
117 self.chromEnd = int(chromEnd)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
118 self.name = name
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
119 self.score = int(score) if score is not None else 0
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
120 self.strand = '-' if str(strand).startswith('-') else '+'
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
121 self.thickStart = self.chromStart
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
122 self.thickEnd = self.chromEnd
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
123 self.itemRgb = '0'
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
124 self.blockCount = int(blockCount)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
125 self.blockSizes = as_int_list(blockSizes)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
126 self.blockStarts = as_int_list(blockStarts)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
127 self.protacc = protacc
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
128 self.peptide = peptide
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
129 self.uniqueness = uniqueness
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
130 self.genomeReference = genomeReference
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
131 self.psmScore = psmScore
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
132 self.fdr = fdr
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
133 self.mods = mods
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
134 self.charge = charge
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
135 self.expMassToCharge = expMassToCharge
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
136 self.calcMassToCharge = calcMassToCharge
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
137 self.psmRank = psmRank
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
138 self.datasetID = datasetID
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
139 self.uri = uri
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
140
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
141 def __str__(self):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
142 return '%s\t%d\t%d\t%s\t%d\t%s\t%d\t%d\t%s\t%d\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n' % \
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
143 (self.chrom, self.chromStart, self.chromEnd,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
144 self.name, self.score, self.strand,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
145 self.thickStart, self.thickEnd, self.itemRgb,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
146 self.blockCount,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
147 ','.join([str(x) for x in self.blockSizes]),
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
148 ','.join([str(x) for x in self.blockStarts]),
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
149 self.protacc, self.peptide, self.uniqueness,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
150 self.genomeReference,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
151 self.psmScore, self.fdr, self.mods,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
152 self.charge, self.expMassToCharge, self.calcMassToCharge,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
153 self.psmRank, self.datasetID, self.uri)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
154
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
155
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
156 class ProBED ( object ):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
157 def __init__(self,species=None,assembly=None,comments=[]):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
158 self.species = species
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
159 self.assembly = assembly
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
160 self.comments = comments
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
161 self.entries = dict()
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
162
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
163 def add_entry(self,entry):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
164 if not entry.chrom in self.entries:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
165 self.entries[entry.chrom] = []
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
166 self.entries[entry.chrom].append(entry)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
167
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
168 def write(self,fh):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
169 rnames = sort_chrom_names(self.entries.keys())
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
170 for sn in rnames:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
171 if sn not in self.entries:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
172 continue
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
173 for pbe in sorted(self.entries[sn], key=attrgetter('chromStart','chromEnd')):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
174 fh.write(str(pbe))
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
175
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
176
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
177 class ProBAMEntry (object):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
178 def __init__(self, qname='', flag=0, rname='', pos=0, mapq=255, cigar='', rnext='*', pnext='0', tlen='0', seq='*', qual='*', optional=PROBAM_DEFAULTS):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
179 self.qname = qname
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
180 self.flag = flag
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
181 self.rname = rname
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
182 self.pos = pos
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
183 self.mapq = mapq
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
184 self.cigar = cigar
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
185 self.rnext = rnext
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
186 self.pnext = pnext
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
187 self.tlen = tlen
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
188 self.seq = seq
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
189 self.qual = qual
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
190 self.optional = optional
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
191 def __str__(self):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
192 opt_cols = '\t%s' % '\t'.join(['%s:%s:%s' % (t,PROBAM_TYTPES[t],self.optional[t]) for t in PROBAM_TAGS]) if self.optional else ''
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
193 return '%s\t%d\t%s\t%d\t%d\t%s\t%s\t%s\t%s\t%s\t%s%s' % (
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
194 self.qname,self.flag,self.rname,self.pos,self.mapq,self.cigar,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
195 str(self.rnext) if self.rnext else '',
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
196 str(self.pnext) if self.pnext else '',
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
197 str(self.tlen) if self.tlen else '',
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
198 self.seq,
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
199 self.qual, opt_cols)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
200 def add_optional(self,tag,value):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
201 self.optional[tag] = value
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
202
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
203
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
204 class ProBAM ( object ):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
205 def __init__(self,species=None,assembly=None,seqlens={},comments=[]):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
206 self.species = species
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
207 self.assembly = assembly
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
208 self.seqlens = seqlens
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
209 self.comments = comments
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
210 self.entries = dict()
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
211 self.opt_columns = set()
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
212 self.rg = []
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
213
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
214 def add_entry(self,pb_entry):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
215 if not pb_entry.rname in self.entries:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
216 self.entries[pb_entry.rname] = []
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
217 self.entries[pb_entry.rname].append(pb_entry)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
218 if pb_entry.optional:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
219 self.opt_columns | set(pb_entry.optional.keys())
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
220
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
221 def add_entry_from_bed(self,bed_entry,optional=dict()):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
222 if bed_entry.pep:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
223 optional['XP:Z'] = bed_entry.pep
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
224 qname=bed_entry.name
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
225 flag = 0 if bed_entry.strand == '+' else 16
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
226 rname = bed_entry.chrom
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
227 pos = bed_entry.chromStart + 1
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
228 cigar = bed_entry.get_cigar()
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
229 seq = bed_entry.get_spliced_seq(strand='+') if bed_entry.seq else '*'
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
230 pb_entry = ProBAMEntry(qname=qname, flag=flag, rname=rname, pos=pos,cigar=cigar,seq=seq,optional=optional)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
231 self.add_entry(pb_entry)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
232
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
233 def write(self,fh):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
234 fh.write('@HD VN:1.0 SO:coordinate\n')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
235 rnames = sort_chrom_names(self.seqlens.keys())
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
236 for sn in rnames:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
237 fh.write('@SQ\tSN:%s\tLN:%d\n' % (sn,self.seqlens[sn]))
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
238 for rg in self.rg:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
239 fh.write('@RG\tID:%s\n' % (rg))
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
240 fh.write('@PG\tID:SampleSpecificGenerator\n')
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
241 for comment in self.comments:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
242 fh.write('@CO\t%s\n' % comment)
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
243 for sn in rnames:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
244 if sn not in self.entries:
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
245 continue
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
246 for pbe in sorted(self.entries[sn], key=attrgetter('pos')):
f2dc9805107a planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/mzsqlite_psm_align commit 464e05be1084ed9a65b542c8eabb18147d425666
galaxyp
parents:
diff changeset
247 fh.write('%s\n' % str(pbe))