Mercurial > repos > public-health-bioinformatics > assign_clades

annotate assign_clades.py @ 0:a971083404a4 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/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
author public-health-bioinformatics
date Thu, 04 Jul 2019 19:34:32 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
1 #!/usr/bin/env python
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
2
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
3 '''Accepts fasta files containing amino acid sequence, reading them in as
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
4 amino acid sequence objects. Reads influenza clade defintions (i.e. amino
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
5 acids at certain positions) from .csv file into dictionary structure. Searches
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
6 each of the amino acid sequence objects for a list of matching clades, assigns
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
7 the most 'evolved' (i.e. child as opposed to parent clade) to the sequence. Appends
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
8 "_cladename" to the Sequence name and generates a fasta file of original sequences with
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
9 modified names.'''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
10
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
11 '''Author: Diane Eisler, Molecular Microbiology & Genomics, BCCDC Public Health Laboratory, Oct 2017'''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
12
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
13 import sys,string,os, time, Bio
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
14 from Bio import Seq, SeqIO, SeqUtils, Alphabet, SeqRecord
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
15 from Bio.SeqRecord import SeqRecord
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
16 from Bio.Alphabet import IUPAC
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
17 from Bio.Seq import Seq
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
18
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
19 localtime = time.asctime(time.localtime(time.time())) #date and time of analysis
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
20 inFileHandle1 = sys.argv[1] #batch fasta file with sequences to be parsed
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
21 inFileHandle2 = sys.argv[2] # .csv file containing clade definitions and "depth"
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
22 outFileHandle = sys.argv[3] #user-specified name for output file of aa seq's with clade suffixes
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
23 outFile= open(outFileHandle,'w') #open a writable, appendable output file
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
24 seqList = [] #list of aa sequence objects to parse for clade definitions
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
25 cladeList = [] #empty list to hold clade tuples i.e. ("3C.3a", 1 ,{"3":"I", "9":"V"..})
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
26
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
27 '''Searches record for required amino acids at defined positions. If found, assigns
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
28 clade name to sequence name by appending underscore and clade name to record id.'''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
29 def call_clade(record):
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
30 print("---------------------------------------------------------------------")
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
31 print("Parsing %s for matching flu clade definitions..." % (record.id))
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
32 matchList = [] #empty list to hold clades that match 100%
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
33 #iterate over each tuple in the clade list
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
34 for clade in cladeList:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
35 cladeName = clade[0] #temp variable for name
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
36 depth = clade[1] #temp variable for depth
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
37 sites = clade[2] #temp variable for aa def dictionary
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
38 shouldFind = len(sites) #number of sites that should match
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
39 found = 0 #a counter to hold matches to antigenic sites
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
40 #iterate over each position in sites dictionary
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
41 for pos, aa in sites.items():
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
42 #translate pos to corresponding index in target sequence
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
43 index = int(pos) - 1
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
44 #if record at index has same amino acid as 'aa', increment 'found'
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
45 if record[index] == aa:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
46 found += 1
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
47 if (found == shouldFind):
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
48 #add the matching clade tuple to the list of matches
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
49 matchList.append(clade)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
50 return matchList
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
51
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
52 '''Compares depth level of clades in a list and returns the most granular one'''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
53 def decide_clade_by_depth(matchList):
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
54 #empty variable for maximum depth encountered
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
55 max_depth = 0
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
56 best_match_name = '' #variable to hold most granular clade
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
57 #for each matching clade, check depth of the corresponding tuple
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
58 for clade in matchList:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
59 #if the current clade is 'deeper' than the one before it
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
60 if clade[1] > max_depth:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
61 #store this depth
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
62 max_depth = clade[1]
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
63 #store name of the clade
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
64 best_match_name = clade[0]
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
65 return best_match_name
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
66
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
67 '''opens the .csv file of clade definitions and clade "depth" '''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
68 with open (inFileHandle2, 'r') as clade_file:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
69 #remove whitespace from the end of each line and split elements at commas
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
70 for line in clade_file:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
71 #print "Current Line in File:" + line
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
72 sites={} #initialize a dictionary for clade
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
73 elementList = line.rstrip().split(',')
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
74 new_list = [] #start a new list to put non-empty strings into
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
75 #remove empty stings in list
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
76 for item in elementList:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
77 if item != '':
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
78 new_list.append(item)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
79 name = new_list.pop(0) #move 1st element to name field
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
80 depth = int(new_list.pop(0)) #move 2nd element to depth field
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
81 #read remaining pairs of non-null elements into clade def dictionary
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
82 for i in range(0, len(new_list), 2):
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
83 #move next 2 items from the list into the dict
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
84 pos = new_list[i]
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
85 aa = new_list[i + 1]
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
86 sites[pos] = aa
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
87 #add the clade info as a tuple to the cladeList[]
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
88 oneClade =(name, depth, sites)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
89 cladeList.append(oneClade)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
90 print("The List of Clades:")
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
91 for clade in cladeList:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
92 print("Clade Name: %s Depth: %i Antigenic Sites: %i" % (clade[0], clade[1], len(clade[2])))
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
93 for pos, aa in clade[2].items():
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
94 print("Pos: %s\tAA: %s" % (pos,aa))
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
95
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
96 '''opens readable input file of sequences to parse using filename from cmd line,
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
97 instantiates as AA Sequence objects, with ppercase sequences'''
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
98 with open(inFileHandle1,'r') as inFile:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
99 #read in Sequences from fasta file, uppercase and add to seqList
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
100 for record in SeqIO.parse(inFile, "fasta", alphabet=IUPAC.protein):
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
101 record = record.upper()
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
102 seqList.append(record) #add Seq to list of Sequences
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
103 print("\n%i flu HA sequences will be compared to current clade definitions..." % len(seqList))
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
104 #parse each target sequence object
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
105 for record in seqList:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
106 clade_call = '' #empty variale for final clade call on sequence
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
107 matchingCladeList = call_clade(record) #holds matching clade tuples
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
108 #if there is more than one clade match
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
109 if len(matchingCladeList) > 1:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
110 #choose the most granular clade based on depth
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
111 clade_call = decide_clade_by_depth(matchingCladeList)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
112 #if there is only one clade call
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
113 elif len(matchingCladeList) > 0:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
114 clade = matchingCladeList[0] #take the first tuple in the list
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
115 clade_call = clade[0] #clade name is the first item in the tuple
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
116 #empty list return, no matches
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
117 else:
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
118 clade_call = "No_Match"
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
119 print(clade_call)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
120 seq_name = record.id
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
121 mod_name = seq_name + "_" + clade_call
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
122 print("New Sequence Name: " + mod_name)
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
123 record.id = mod_name
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
124
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
125
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
126 #output fasta file with clade calls appended to sequence names
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
127 SeqIO.write(seqList,outFile,"fasta")
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
128
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
129 #print("\n%i Sequences Extracted to Output file: %s" % ((len(extractedSeqList),outFileHandle)))
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
130 inFile.close()
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
131 clade_file.close()
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
132 outFile.close()
a971083404a4 planemo upload for repository https://github.com/Public-Health-Bioinformatics/flu_classification_suite commit b96b6e06f6eaa6ae8ef4c24630dbb72a4aed7dbe
public-health-bioinformatics
parents:
diff changeset
133