Mercurial > repos > charles_s_test > seqsero2
annotate run_seqsero.py @ 0:6895de35a263 draft
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
author | charles_s_test |
---|---|
date | Thu, 19 Oct 2017 18:16:51 -0400 |
parents | |
children | 38ad1130d077 |
rev | line source |
---|---|
0
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
1 #!/usr/bin/python |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
2 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
3 import os, re, sys, time, datetime |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
4 import subprocess |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
5 from subprocess import Popen, PIPE |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
6 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
7 database = '/nfs/sw/apps/galaxy/database/universe.sqlite' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
8 seqsero = '/nfs/sw/apps/galaxy/tools/seqsero/SeqSero.py' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
9 test_out = open("/nfs/sw/apps/galaxy/tools/seqsero/test.txt", 'w') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
10 out_path = '/nfs/sw/apps/galaxy/database/files/000' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
11 test_out2 = open("/nfs/sw/apps/galaxy/tools/seqsero/test2.txt", 'w') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
12 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
13 path2sample = '/nfs/sw/apps/galaxy/tools/seqsero/fastq_tmp' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
14 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
15 #test_out2.write(" Test test test"); |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
16 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
17 test_out.write("monkey "); |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
18 #test_out.write(""); |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
19 test_out.write("\t".join(sys.argv)+'\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
20 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
21 fq_list1 = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
22 fq_list2 = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
23 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
24 test_out.write(str(len(sys.argv))+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
25 if len(sys.argv) >= 2: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
26 test_out.write("\t".join(sys.argv)+'\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
27 fq_list1 = sys.argv[1] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
28 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
29 test_out.write(str(len(sys.argv))+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
30 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
31 fastq_files = re.split(",", fq_list1) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
32 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
33 def print_time(): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
34 test_out.write(time.asctime( time.localtime(time.time()))) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
35 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
36 tmp_path = "/nfs/sw/apps/galaxy/tools/seqsero/fastq_tmp" |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
37 if not os.path.exists(tmp_path): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
38 os.system('mkdir '+tmp_path) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
39 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
40 test_out.write(str(len(fastq_files))+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
41 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
42 def list_runs(fastq_files): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
43 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
44 Creates dict with runs as keys and list with filenames as values. |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
45 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
46 # print "monkey" |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
47 run2fastqs = {} |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
48 for file in fastq_files: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
49 run = '' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
50 try: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
51 fastq = open(file, 'r') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
52 i = 0 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
53 for line in fastq: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
54 line = line.rstrip("\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
55 if i == 0: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
56 run = re.split("\s", line)[0] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
57 run = re.sub('@', '', run) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
58 run = re.sub("\.1", '', run) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
59 else: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
60 break |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
61 i += 1 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
62 file1 = re.split('/', file)[-1] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
63 file2 = re.sub('.dat$', '_'+run+'.fastq', file1) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
64 new_path_file = tmp_path+'/'+file2 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
65 # print run, new_path_file |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
66 if file2 not in os.listdir(tmp_path): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
67 os.system('cp '+file+' '+new_path_file) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
68 if run in run2fastqs.keys(): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
69 if file not in run2fastqs[run]: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
70 run2fastqs[run].append(new_path_file) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
71 else: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
72 run2fastqs[run] = [new_path_file] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
73 except IOError: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
74 print "Data not found. It is possible for a deleted file to still be listed "\ |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
75 "in a Galaxy library. Please confirm that the data still exists on this "\ |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
76 "server. You may need to upload it again." |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
77 return run2fastqs |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
78 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
79 def run_seqsero(run2fastqs): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
80 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
81 Takes files from run2fastqs and runs SeqSero. |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
82 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
83 outputs = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
84 for run in run2fastqs: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
85 seqsero_cmd = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
86 if len(run2fastqs[run]) == 2: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
87 seqsero_cmd = ['python', seqsero, '-m', '2', '-i', run2fastqs[run][0], run2fastqs[run][1]] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
88 elif len(run2fastqs[run]) == 1: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
89 seqsero_cmd = ['python', seqsero, '-m', '1', '-i', run2fastqs[run][0]] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
90 p = Popen(seqsero_cmd, stdout=PIPE) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
91 output = p.communicate() |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
92 outputs.append(output) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
93 return outputs |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
94 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
95 def get_serotypes(outputs): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
96 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
97 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
98 fastq2comment = {} |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
99 fastq2serotype = {} |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
100 for sample in outputs: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
101 fastqs = '' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
102 lines_used = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
103 for line in sample: # line is actually the entire seqsero output. |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
104 line = str(line) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
105 linel = re.split("\n", line) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
106 # print linel |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
107 #lines_used = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
108 for element in linel: # element is a line of seqsero output. |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
109 element = element.rstrip("\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
110 test_out.write(element+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
111 elementl = re.split("\t", element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
112 if elementl[0] == 'Input files:': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
113 fastqs = elementl[1] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
114 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
115 if elementl[1] not in fastq2serotype.keys(): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
116 fastq2serotype[fastqs] = ['']*5 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
117 fastq2comment[fastqs] = [] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
118 elif elementl[0] == 'O antigen prediction:': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
119 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
120 fastq2serotype[fastqs][0] = elementl[1] # add predicted profile |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
121 elif elementl[0] == 'H1 antigen prediction(fliC):': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
122 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
123 fastq2serotype[fastqs][1] = elementl[1] # add predicted profile |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
124 elif elementl[0] == 'H2 antigen prediction(fljB):': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
125 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
126 fastq2serotype[fastqs][2] = elementl[1] # add predicted profile |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
127 elif elementl[0] == 'Predicted antigenic profile:': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
128 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
129 fastq2serotype[fastqs][3] = elementl[1] # add predicted profile |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
130 elif elementl[0] == 'Predicted serotype(s):': |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
131 lines_used.append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
132 fastq2serotype[fastqs][4] = elementl[1] # add predicted serotype |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
133 if element not in lines_used and re.search("\w", fastqs) and len(element) > 7: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
134 fastq2comment[fastqs].append(element) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
135 #print "\n" |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
136 return fastq2serotype, fastq2comment |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
137 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
138 def print_html(fastq2serotype, fastq2comment): |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
139 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
140 Takes dict and prints to html file. |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
141 ''' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
142 tab_out = open('Seqsero_result.txt', 'w'); |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
143 html_out = open('Seqsero_result.html', 'w') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
144 html_out.write('<!DOCTYPE html>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
145 html_out.write('<html>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
146 html_out.write('<head>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
147 html_out.write('<title>SeqSero Results</title>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
148 html_out.write('</head>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
149 html_out.write('<body>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
150 html_out.write('<body style="font-family:Helvetica;">\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
151 html_out.write('<p style="font-size:10px">\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
152 html_out.write('<table border=1>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
153 header = 'Input Files\tO antigen prediction\tH1 antigen prediction(fliC)\tH2 antigen prediction(fljB)\tPredicted antigenic profile\tPredicted serotype(s)' |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
154 header = re.sub(' ', '_', header) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
155 header_l = ['Input Files', 'O antigen prediction', 'H1 antigen prediction(fliC)', 'H2 antigen prediction(fljB)', 'Predicted antigenic profile', 'Predicted serotype(s)'] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
156 html_out.write('<tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
157 for element in header_l: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
158 html_out.write('<td>'+element+'</td>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
159 html_out.write('</tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
160 tab_out.write(header+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
161 print "\n\n", header |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
162 for fastq in fastq2serotype: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
163 # print fastq, fastq2serotype[fastq] |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
164 line_to_print = fastq+'\t'+"\t".join(fastq2serotype[fastq]) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
165 tab_out.write(line_to_print+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
166 html_out.write('<tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
167 html_out.write('<td>'+fastq+'</td>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
168 for antigen in fastq2serotype[fastq]: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
169 html_out.write('<td>'+antigen+'</td>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
170 html_out.write('</tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
171 print line_to_print |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
172 html_out.write('</table>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
173 print "\n" |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
174 for fastq in fastq2comment: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
175 tab_out.write("\n"+fastq+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
176 html_out.write('<tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
177 html_out.write('<p>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
178 html_out.write('<td>'+fastq+"<br></td>\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
179 for line in fastq2comment[fastq]: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
180 #if len(line) > 7: |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
181 html_out.write('<td>'+line+'</td>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
182 tab_out.write(line+"\n") |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
183 print line |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
184 print "\n" |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
185 html_out.write('</p>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
186 html_out.write('</tr>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
187 html_out.write('</body>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
188 html_out.write('</html>\n') |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
189 html_out.close() |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
190 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
191 run2fastqs = list_runs(fastq_files) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
192 outputs = run_seqsero(run2fastqs) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
193 fastq2serotype, fastq2comment = get_serotypes(outputs) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
194 print_html(fastq2serotype, fastq2comment) |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
195 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
196 print_time() |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
197 |
6895de35a263
planemo upload commit 844a891e4eaf732830043204ac636907eefb011d-dirty
charles_s_test
parents:
diff
changeset
|
198 |