cpt_find_spanins: findSpanin.py annotate

annotate findSpanin.py @ 4:673d1776d3b9 draft

planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc

author	cpt
date	Fri, 05 Jan 2024 05:51:06 +0000
parents	fd70980a516b
children	46b252c89e9e

rev	line source
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	1 ##### findSpanin.pl --> findSpanin.py
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	2 ######### Much of this code is very "blocked", in the sense that one thing happens...then a function happens on the return...then another function...etc...etc...
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	3
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	4 import argparse
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	5 import os
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	6 import re # new
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	7 import itertools # new
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	8 from collections import Counter, OrderedDict
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	9 from spaninFuncs import (
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	10 getDescriptions,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	11 grabLocs,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	12 spaninProximity,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	13 splitStrands,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	14 tuple_fasta,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	15 lineWrapper,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	16 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	17
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	18 ### Requirement Inputs
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	19 #### INPUT : putative_isp.fa & putative_osp.fa (in that order)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	20 #### PARAMETERS :
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	21
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	22
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	23 ###############################################################################
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	24 def write_output(candidates):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	25 """output file function...maybe not needed"""
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	26 pass
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	27
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	28
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	29 def reconfigure_dict(spanins):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	30 """
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	31 re organizes dictionary to be more friendly for checks
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	32 """
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	33
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	34 new_spanin_dict = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	35
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	36 for each_spanin_type, data_dict in spanins.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	37 # print(f"{each_spanin_type} == {data_dict}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	38 new_spanin_dict[each_spanin_type] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	39 new_spanin_dict[each_spanin_type]["positive"] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	40 new_spanin_dict[each_spanin_type]["negative"] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	41 new_spanin_dict[each_spanin_type]["positive"]["coords"] = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	42 new_spanin_dict[each_spanin_type]["negative"]["coords"] = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	43 for outter_orf, inner_data in data_dict.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	44 list_of_hits = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	45 for data_content in inner_data:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	46 # print(data_content)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	47 data_content.insert(0, outter_orf)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	48 # print(f"new data_content -> {data_content}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	49 # print(data_content)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	50 # list_of_hits += [data_content]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	51 # new_spanin_dict[each_spanin_type] += [data_content]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	52 if data_content[6] == "+":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	53 # print(f"{each_spanin_type} @ POSITIVE")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	54 new_spanin_dict[each_spanin_type]["positive"]["coords"] += [
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	55 data_content
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	56 ]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	57 elif data_content[6] == "-":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	58 # print(f"{each_spanin_type} @ NEGATIVE")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	59 new_spanin_dict[each_spanin_type]["negative"]["coords"] += [
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	60 data_content
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	61 ]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	62 # print(new_spanin_dict[each_spanin_type])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	63 # print(reorganized)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	64 # print(f"{outter_orf} => {inner_data}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	65 # print(new_spanin_dict)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	66
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	67 # print('\n')
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	68 # for k, v in new_spanin_dict.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	69 # print(k)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	70 # print(v)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	71 return new_spanin_dict
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	72
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	73
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	74 def check_for_uniques(spanins):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	75 """
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	76 Checks for unique spanins based on spanin_type.
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	77 If the positive strand end site is _the same_ for a i-spanin, we would group that as "1".
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	78 i.e. if ORF1, ORF2, and ORF3 all ended with location 4231, they would not be unique.
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	79 """
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	80 pair_dict = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	81 pair_dict = {
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	82 "pairs": {
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	83 "location_amount": [],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	84 "pair_number": {},
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	85 }
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	86 }
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	87 for each_spanin_type, spanin_data in spanins.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	88 # print(f"{each_spanin_type} ===> {spanin_data}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	89 # early declarations for cases of no results
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	90 pos_check = [] # end checks
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	91 pos_uniques = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	92 neg_check = [] # start checks
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	93 neg_uniques = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	94 unique_ends = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	95 pos_amt_unique = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	96 neg_amt_unique = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	97 amt_positive = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	98 amt_negative = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	99 spanin_data["uniques"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	100 spanin_data["amount"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	101 # spanin_data['positive']['amt_positive'] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	102 # spanin_data['positive']['pos_amt_unique'] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	103 # spanin_data['positive']['isp_match'] = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	104 # spanin_data['negative']['amt_negative'] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	105 # spanin_data['negative']['neg_amt_unique'] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	106 # spanin_data['negative']['isp_match'] = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	107 # print(spanin_data)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	108 if spanin_data["positive"]["coords"]:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	109 # do something...
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	110 # print('in other function')
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	111 # print(spanin_data['positive']['coords'])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	112 for each_hit in spanin_data["positive"]["coords"]:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	113 pos_check.append(each_hit[2])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	114 pair_dict["pairs"]["location_amount"].append(each_hit[2])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	115 pos_uniques = list(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	116 set(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	117 [
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	118 end_site
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	119 for end_site in pos_check
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	120 if pos_check.count(end_site) >= 1
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	121 ]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	122 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	123 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	124 # print(pos_check)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	125 # print(pos_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	126 amt_positive = len(spanin_data["positive"]["coords"])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	127 pos_amt_unique = len(pos_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	128 if amt_positive:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	129 spanin_data["positive"]["amt_positive"] = amt_positive
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	130 spanin_data["positive"]["pos_amt_unique"] = pos_amt_unique
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	131 # pair_dict['pairs']['locations'].extend(pos_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	132 else:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	133 spanin_data["positive"]["amt_positive"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	134 spanin_data["positive"]["pos_amt_unique"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	135 if spanin_data["negative"]["coords"]:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	136 # do something else...
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	137 # print('in other function')
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	138 # print(spanin_data['negative']['coords'])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	139 for each_hit in spanin_data["negative"]["coords"]:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	140 neg_check.append(each_hit[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	141 pair_dict["pairs"]["location_amount"].append(each_hit[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	142 neg_uniques = list(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	143 set(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	144 [
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	145 start_site
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	146 for start_site in neg_check
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	147 if neg_check.count(start_site) >= 1
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	148 ]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	149 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	150 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	151 # print(neg_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	152 amt_negative = len(spanin_data["negative"]["coords"])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	153 neg_amt_unique = len(neg_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	154 if amt_negative:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	155 spanin_data["negative"]["amt_negative"] = amt_negative
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	156 spanin_data["negative"]["neg_amt_unique"] = neg_amt_unique
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	157 # pair_dict['pairs']['locations'].extend(neg_uniques)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	158 else:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	159 spanin_data["negative"]["amt_negative"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	160 spanin_data["negative"]["neg_amt_unique"] = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	161 spanin_data["uniques"] += (
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	162 spanin_data["positive"]["pos_amt_unique"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	163 + spanin_data["negative"]["neg_amt_unique"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	164 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	165 spanin_data["amount"] += (
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	166 spanin_data["positive"]["amt_positive"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	167 + spanin_data["negative"]["amt_negative"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	168 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	169 # print(spanin_data['uniques'])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	170 list(set(pair_dict["pairs"]["location_amount"]))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	171 pair_dict["pairs"]["location_amount"] = dict(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	172 Counter(pair_dict["pairs"]["location_amount"])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	173 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	174 for data in pair_dict.values():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	175 # print(data['locations'])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	176 # print(type(data['locations']))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	177 v = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	178 for loc, count in data["location_amount"].items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	179 # data['pair_number'] = {loc
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	180 v += 1
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	181 data["pair_number"][loc] = v
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	182 # print(dict(Counter(pair_dict['pairs']['locations'])))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	183 # print(pair_dict)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	184 spanins["total_amount"] = (
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	185 spanins["EMBEDDED"]["amount"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	186 + spanins["SEPARATED"]["amount"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	187 + spanins["OVERLAPPED"]["amount"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	188 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	189 spanins["total_unique"] = (
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	190 spanins["EMBEDDED"]["uniques"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	191 + spanins["SEPARATED"]["uniques"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	192 + spanins["OVERLAPPED"]["uniques"]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	193 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	194 # spanins['total_unique'] = len(pair_dict['pairs']['pair_number'])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	195 return spanins, pair_dict
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	196
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	197
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	198 if __name__ == "__main__":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	199 # Common parameters for both ISP / OSP portion of script
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	200
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	201 parser = argparse.ArgumentParser(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	202 description="Trim the putative protein candidates and find potential i-spanin / o-spanin pairs"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	203 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	204
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	205 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	206 "putative_isp_fasta_file",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	207 type=argparse.FileType("r"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	208 help='Putative i-spanin FASTA file, output of "generate-putative-isp"',
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	209 ) # the "input" argument
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	210
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	211 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	212 "putative_osp_fasta_file",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	213 type=argparse.FileType("r"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	214 help='Putative o-spanin FASTA file, output of "generate-putative-osp"',
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	215 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	216
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	217 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	218 "--max_isp_osp_distance",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	219 dest="max_isp_osp_distance",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	220 default=10,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	221 type=int,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	222 help="max distance from end of i-spanin to start of o-spanin, measured in AAs",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	223 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	224
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	225 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	226 "--embedded_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	227 dest="embedded_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	228 type=argparse.FileType("w"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	229 default="_findSpanin_embedded_results.txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	230 help="Results of potential embedded spanins",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	231 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	232 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	233 "--overlap_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	234 dest="overlap_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	235 type=argparse.FileType("w"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	236 default="_findSpanin_overlap_results.txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	237 help="Results of potential overlapping spanins",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	238 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	239 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	240 "--separate_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	241 dest="separate_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	242 type=argparse.FileType("w"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	243 default="_findSpanin_separated_results.txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	244 help="Results of potential separated spanins",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	245 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	246
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	247 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	248 "--summary_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	249 dest="summary_txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	250 type=argparse.FileType("w"),
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	251 default="_findSpanin_summary.txt",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	252 help="Results of potential spanin pairs",
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	253 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	254 parser.add_argument(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	255 "-v", action="version", version="0.3.0"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	256 ) # Is this manually updated?
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	257 args = parser.parse_args()
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	258
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	259 #### RE-WRITE
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	260 SPANIN_TYPES = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	261 SPANIN_TYPES["EMBEDDED"] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	262 SPANIN_TYPES["OVERLAPPED"] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	263 SPANIN_TYPES["SEPARATED"] = {}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	264 # SPANIN_TYPES = {
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	265 # 'EMBEDDED' : {},
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	266 # 'OVERLAPPED' : {},
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	267 # 'SEPARATED' : {},
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	268 # }
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	269
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	270 isp = getDescriptions(args.putative_isp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	271 args.putative_isp_fasta_file = open(args.putative_isp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	272 isp_full = tuple_fasta(args.putative_isp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	273
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	274 osp = getDescriptions(args.putative_osp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	275 args.putative_osp_fasta_file = open(args.putative_osp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	276 osp_full = tuple_fasta(args.putative_osp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	277
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	278 #### location data
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	279 location_data = {"isp": [], "osp": []}
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	280 spanins = [isp, osp]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	281 for idx, each_spanin_type in enumerate(spanins):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	282 for description in each_spanin_type:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	283 locations = grabLocs(description)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	284 if idx == 0: # i-spanin
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	285 location_data["isp"].append(locations)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	286 elif idx == 1: # o-spanin
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	287 location_data["osp"].append(locations)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	288
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	289 #### Check for types of spanins
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	290 embedded, overlap, separate = spaninProximity(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	291 isp=location_data["isp"],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	292 osp=location_data["osp"],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	293 max_dist=args.max_isp_osp_distance * 3,
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	294 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	295
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	296 SPANIN_TYPES["EMBEDDED"] = embedded
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	297 SPANIN_TYPES["OVERLAPPED"] = overlap
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	298 SPANIN_TYPES["SEPARATED"] = separate
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	299
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	300 # for spanin_type, spanin in SPANIN_TYPES.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	301 # s = 0
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	302 # for sequence in spanin.values():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	303 # s += len(sequence)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	304 # SPANIN_TYPES[spanin_type]['amount'] = s
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	305 # SPANIN_TYPES[spanin_type]['unique'] = len(spanin.keys())
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	306
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	307 # check_for_unique_spanins(SPANIN_TYPES)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	308 spanins = reconfigure_dict(SPANIN_TYPES)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	309 spanins, pair_dict = check_for_uniques(spanins)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	310 # print(pair_dict)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	311 with args.summary_txt as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	312 for each_spanin_type, spanin_data in spanins.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	313 try:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	314 if each_spanin_type not in ["total_amount", "total_unique"]:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	315 # print(each_spanin_type)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	316 # print(each_spanin_type)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	317 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	318 "=~~~~~= "
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	319 + str(each_spanin_type)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	320 + " Spanin Candidate Statistics =~~~~~=\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	321 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	322 f.writelines(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	323 "Total Candidate Pairs = " + str(spanin_data["amount"]) + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	324 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	325 f.writelines(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	326 "Total Unique Pairs = " + str(spanin_data["uniques"]) + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	327 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	328 if each_spanin_type == "EMBEDDED":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	329 for k, v in SPANIN_TYPES["EMBEDDED"].items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	330 # print(k)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	331 f.writelines(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	332 +str(k)
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	333 + " ==> Amount of corresponding candidate o-spanins(s): "
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	334 + str(len(v))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	335 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	336 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	337 if each_spanin_type == "SEPARATED":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	338 for k, v in SPANIN_TYPES["SEPARATED"].items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	339 f.writelines(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	340 +str(k)
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	341 + " ==> Amount of corresponding candidate o-spanins(s): "
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	342 + str(len(v))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	343 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	344 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	345 if each_spanin_type == "OVERLAPPED":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	346 for k, v in SPANIN_TYPES["OVERLAPPED"].items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	347 f.writelines(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	348 +str(k)
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	349 + " ==> Amount of corresponding candidate o-spanins(s): "
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	350 + str(len(v))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	351 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	352 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	353 except TypeError:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	354 continue
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	355 f.write("\n=~~~~~= Tally from ALL spanin types =~~~~~=\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	356 f.writelines("Total Candidates = " + str(spanins["total_amount"]) + "\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	357 f.writelines(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	358 "Total Unique Candidate Pairs = " + str(spanins["total_unique"]) + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	359 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	360
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	361 args.putative_isp_fasta_file = open(args.putative_isp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	362 isp_full = tuple_fasta(args.putative_isp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	363
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	364 args.putative_osp_fasta_file = open(args.putative_osp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	365 osp_full = tuple_fasta(args.putative_osp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	366
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	367 # print(isp_full)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	368 isp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	369 osp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	370 for isp_tupe in isp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	371 # print(isp_tupe)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	372 for pisp, posp in embedded.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	373 # print(f"ISP = searching for {pisp} in {isp_tupe[0]}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	374 if re.search(("(" + str(pisp) + ")\D"), isp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	375 # print(isp_tupe[0])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	376 # print(peri_count)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	377 peri_count = str.split(isp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	378 isp_seqs.append((pisp, isp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	379 # print(isp_seqs)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	380 for osp_tupe in osp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	381 for pisp, posp in embedded.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	382 for data in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	383 # print(f"OSP = searching for {data[3]} in {osp_tupe[0]}, coming from this object: {data}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	384 if re.search(("(" + str(data[3]) + ")\D"), osp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	385 peri_count = str.split(osp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	386 osp_seqs.append((data[3], osp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	387
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	388 with args.embedded_txt as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	389 f.write("================ embedded spanin candidates =================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	390 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	391 "isp\tisp_start\tisp_end\tosp\tosp_start\tosp_end\tstrand\tpair_number\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	392 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	393 if embedded != {}:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	394 # print(embedded)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	395 for pisp, posp in embedded.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	396 # print(f"{pisp} - {posp}")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	397 f.write(pisp + "\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	398 for each_posp in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	399 # print(posp)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	400 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	401 "\t{}\t{}\t{}\t{}\t{}\t{}\t".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	402 each_posp[1],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	403 each_posp[2],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	404 each_posp[3],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	405 each_posp[4],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	406 each_posp[5],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	407 each_posp[6],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	408 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	409 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	410 if each_posp[6] == "+":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	411 if each_posp[2] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	412 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	413 +str(pair_dict["pairs"]["pair_number"][each_posp[2]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	414 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	415 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	416 elif each_posp[6] == "-":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	417 if each_posp[1] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	418 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	419 +str(pair_dict["pairs"]["pair_number"][each_posp[1]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	420 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	421 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	422 else:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	423 f.write("nothing found")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	424
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	425 with open(args.embedded_txt.name, "a") as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	426 f.write("\n================= embedded candidate sequences ================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	427 f.write("======================= isp ==========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	428 for isp_data in isp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	429 # print(isp_data)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	430 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	431 ">isp_orf::{}-peri_count~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	432 isp_data[0], isp_data[2], lineWrapper(isp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	433 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	434 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	435 f.write("\n======================= osp ========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	436 for osp_data in osp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	437 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	438 ">osp_orf::{}-peri_count~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	439 osp_data[0], osp_data[2], lineWrapper(osp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	440 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	441 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	442
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	443 args.putative_isp_fasta_file = open(args.putative_isp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	444 isp_full = tuple_fasta(args.putative_isp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	445
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	446 args.putative_osp_fasta_file = open(args.putative_osp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	447 osp_full = tuple_fasta(args.putative_osp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	448
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	449 isp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	450 osp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	451 for isp_tupe in isp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	452 peri_count = str.split(isp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	453 for pisp, posp in overlap.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	454 if re.search(("(" + str(pisp) + ")\D"), isp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	455 peri_count = str.split(isp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	456 isp_seqs.append((pisp, isp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	457
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	458 for osp_tupe in osp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	459 for pisp, posp in overlap.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	460 for data in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	461 if re.search(("(" + str(data[3]) + ")\D"), osp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	462 peri_count = str.split(osp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	463 osp_seqs.append((data[3], osp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	464
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	465 with args.overlap_txt as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	466 f.write("================ overlap spanin candidates =================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	467 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	468 "isp\tisp_start\tisp_end\tosp\tosp_start\tosp_end\tstrand\tpair_number\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	469 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	470 if overlap != {}:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	471 for pisp, posp in overlap.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	472 f.write(pisp + "\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	473 for each_posp in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	474 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	475 "\t{}\t{}\t{}\t{}\t{}\t{}\t".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	476 each_posp[1],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	477 each_posp[2],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	478 each_posp[3],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	479 each_posp[4],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	480 each_posp[5],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	481 each_posp[6],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	482 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	483 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	484 if each_posp[6] == "+":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	485 if each_posp[2] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	486 # print('ovl ; +')
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	487 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	488 +str(pair_dict["pairs"]["pair_number"][each_posp[2]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	489 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	490 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	491 elif each_posp[6] == "-":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	492 if each_posp[1] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	493 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	494 +str(pair_dict["pairs"]["pair_number"][each_posp[1]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	495 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	496 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	497 else:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	498 f.write("nothing found")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	499
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	500 with open(args.overlap_txt.name, "a") as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	501 # print(isp_seqs)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	502 f.write("\n================= overlap candidate sequences ================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	503 f.write("======================= isp ==========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	504 for isp_data in isp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	505 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	506 ">isp_orf::{}-pericount~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	507 isp_data[0], isp_data[2], lineWrapper(isp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	508 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	509 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	510 f.write("\n======================= osp ========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	511 for osp_data in osp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	512 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	513 ">osp_orf::{}-pericount~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	514 osp_data[0], osp_data[2], lineWrapper(osp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	515 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	516 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	517
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	518 args.putative_isp_fasta_file = open(args.putative_isp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	519 isp_full = tuple_fasta(args.putative_isp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	520 args.putative_osp_fasta_file = open(args.putative_osp_fasta_file.name, "r")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	521 osp_full = tuple_fasta(args.putative_osp_fasta_file)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	522
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	523 isp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	524 osp_seqs = []
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	525 for isp_tupe in isp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	526 for pisp, posp in separate.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	527 if re.search(("(" + str(pisp) + ")\D"), isp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	528 peri_count = str.split(isp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	529 isp_seqs.append((pisp, isp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	530 # print(isp_seqs)
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	531 for osp_tupe in osp_full:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	532 for pisp, posp in separate.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	533 for data in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	534 if re.search(("(" + str(data[3]) + ")\D"), osp_tupe[0]):
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	535 peri_count = str.split(osp_tupe[0], "~=")[1]
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	536 osp_seqs.append((data[3], osp_tupe[1], peri_count))
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	537
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	538 with args.separate_txt as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	539 f.write("================ separated spanin candidates =================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	540 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	541 "isp\tisp_start\tisp_end\tosp\tosp_start\tosp_end\tstrand\tpair_number\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	542 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	543 if separate != {}:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	544 for pisp, posp in separate.items():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	545 f.write(pisp + "\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	546 for each_posp in posp:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	547 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	548 "\t{}\t{}\t{}\t{}\t{}\t{}\t".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	549 each_posp[1],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	550 each_posp[2],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	551 each_posp[3],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	552 each_posp[4],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	553 each_posp[5],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	554 each_posp[6],
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	555 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	556 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	557 if each_posp[6] == "+":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	558 if each_posp[2] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	559 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	560 +str(pair_dict["pairs"]["pair_number"][each_posp[2]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	561 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	562 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	563 elif each_posp[6] == "-":
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	564 if each_posp[1] in pair_dict["pairs"]["pair_number"].keys():
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	565 f.write(
4 673d1776d3b9 planemo upload commit f33bdf952d796c5d7a240b132af3c4cbd102decc cpt parents: 3 diff changeset	566 +str(pair_dict["pairs"]["pair_number"][each_posp[1]])
3 fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	567 + "\n"
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	568 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	569 else:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	570 f.write("nothing found")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	571
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	572 with open(args.separate_txt.name, "a") as f:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	573 f.write("\n================= separated candidate sequences ================\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	574 f.write("======================= isp ==========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	575 for isp_data in isp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	576 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	577 ">isp_orf::{}-pericount~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	578 isp_data[0], isp_data[2], lineWrapper(isp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	579 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	580 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	581 f.write("\n======================= osp ========================\n\n")
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	582 for osp_data in osp_seqs:
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	583 f.write(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	584 ">osp_orf::{}-pericount~={}\n{}\n".format(
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	585 osp_data[0], osp_data[2], lineWrapper(osp_data[1])
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	586 )
fd70980a516b planemo upload commit 94b0cd1fff0826c6db3e7dc0c91c0c5a8be8bb0c cpt parents: diff changeset	587 )

Mercurial > repos > cpt > cpt_find_spanins

annotate findSpanin.py @ 4:673d1776d3b9 draft