ID   CU098_000001
FT   DOMAIN        1    361       NON CYTOPLASMIC.
//
ID   CU098_000002
FT   DOMAIN        1    689       NON CYTOPLASMIC.
//
ID   CU098_000003
FT   DOMAIN        1     23       CYTOPLASMIC.
FT   TRANSMEM     24     45
FT   DOMAIN       46     64       NON CYTOPLASMIC.
FT   TRANSMEM     65     87
FT   DOMAIN       88     98       CYTOPLASMIC.
FT   TRANSMEM     99    121
FT   DOMAIN      122    132       NON CYTOPLASMIC.
FT   TRANSMEM    133    153
FT   DOMAIN      154    164       CYTOPLASMIC.
FT   TRANSMEM    165    182
FT   DOMAIN      183    193       NON CYTOPLASMIC.
FT   TRANSMEM    194    218
FT   DOMAIN      219    236       CYTOPLASMIC.
//
ID   CU098_000004
FT   DOMAIN        1   2847       NON CYTOPLASMIC.
//
ID   CU098_000005
FT   DOMAIN        1    372       NON CYTOPLASMIC.
//
ID   CU098_000006
FT   DOMAIN        1    547       NON CYTOPLASMIC.
//
ID   CU098_000007
FT   DOMAIN        1   1018       NON CYTOPLASMIC.
//
ID   CU098_000008
FT   DOMAIN        1    800       NON CYTOPLASMIC.
//
ID   CU098_000009
FT   DOMAIN        1    783       NON CYTOPLASMIC.
//
ID   CU098_000010
FT   DOMAIN        1    220       CYTOPLASMIC.
FT   TRANSMEM    221    247
FT   DOMAIN      248    252       NON CYTOPLASMIC.
FT   TRANSMEM    253    272
FT   DOMAIN      273    325       CYTOPLASMIC.
FT   TRANSMEM    326    352
FT   DOMAIN      353    363       NON CYTOPLASMIC.
FT   TRANSMEM    364    384
FT   DOMAIN      385    410       CYTOPLASMIC.
FT   TRANSMEM    411    431
FT   DOMAIN      432    461       NON CYTOPLASMIC.
FT   TRANSMEM    462    484
FT   DOMAIN      485    604       CYTOPLASMIC.
FT   TRANSMEM    605    622
FT   DOMAIN      623    633       NON CYTOPLASMIC.
FT   TRANSMEM    634    662
FT   DOMAIN      663    739       CYTOPLASMIC.
//
ID   CU098_000011
FT   DOMAIN        1    338       NON CYTOPLASMIC.
//
ID   CU098_000012
FT   DOMAIN        1     51       CYTOPLASMIC.
FT   TRANSMEM     52     69
FT   DOMAIN       70     71       NON CYTOPLASMIC.
//
ID   CU098_000013
FT   DOMAIN        1    623       NON CYTOPLASMIC.
//
ID   CU098_000014
FT   DOMAIN        1    140       NON CYTOPLASMIC.
//
ID   CU098_000015
FT   DOMAIN        1    103       NON CYTOPLASMIC.
//
ID   CU098_000016
FT   DOMAIN        1    455       NON CYTOPLASMIC.
//
ID   CU098_000017
FT   DOMAIN        1    575       NON CYTOPLASMIC.
//
ID   CU098_000018
FT   DOMAIN        1    342       NON CYTOPLASMIC.
//
ID   CU098_000019
FT   DOMAIN        1      6       CYTOPLASMIC.
FT   TRANSMEM      7     24
FT   DOMAIN       25     43       NON CYTOPLASMIC.
FT   TRANSMEM     44     63
def parsePhobiusSignalP(phobius, sigP, membrane_annot, secretome_annot):
FT   DOMAIN       64     74       CYTOPLASMIC.
FT   TRANSMEM     75     94
FT   DOMAIN       95    105       NON CYTOPLASMIC.
FT   TRANSMEM    106    126
FT   DOMAIN      127    137       CYTOPLASMIC.

�
from __future__ import division
import os, subprocess, logging, sys, argparse, inspect, csv, time, re, shutil, datetime, glob, platform, multiprocessing, itertools, hashlib, math, types
from natsort import natsorted
import warnings
from Bio import SeqIO
with warnings.catch_warnings():
    warnings.simplefilter('ignore')
    from Bio import SearchIO
from Bio import BiopythonWarning
warnings.simplefilter('ignore', BiopythonWarning)

#get the working directory, so you can move back into DB folder to find the files you need
currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))
parentdir = os.path.dirname(currentdir)
sys.path.insert(0,parentdir)
DB = os.path.join(parentdir, 'DB')
LIB = os.path.join(parentdir, 'lib')
UTIL = os.path.join(parentdir, 'util')
GeneMark2GFF = os.path.join(UTIL, 'genemark_gtf2gff3.pl')

pref_colors=["#CF3C57","#65B23A","#6170DD","#D18738","#D542B5",
"#724A63","#60AABA","#5DB07C","#6C5824","#D74B2B","#6B97D6","#893B2E",
"#B68DB7","#564E91","#ACA13C","#3C6171","#436B33","#D84088",
"#D67A77","#9D55C4","#8B336E","#DA77B9","#D850E5","#B188DF"]

Nogs = {'NOG': 'All organisms (5.0GB)',
'aciNOG': 'Acidobacteria (125.3MB)',
'acidNOG': 'Acidobacteriia (75.4MB)',
:1136
FT   TRANSMEM    138    161
FT   DOMAIN      162    172       NON CYTOPLASMIC.
FT   TRANSMEM    173    190
FT   DOMAIN      191    256       CYTOPLASMIC.
FT   TRANSMEM    257    278
FT   DOMAIN      279    292       NON CYTOPLASMIC.
FT   TRANSMEM    293    314
FT   DOMAIN      315    320       CYTOPLASMIC.
FT   TRANSMEM    321    340
FT   DOMAIN      341    351       NON CYTOPLASMIC.
FT   TRANSMEM    352    376
FT   DOMAIN      377    387       CYTOPLASMIC.
FT   TRANSMEM    388    407
FT   DOMAIN      408    418       NON CYTOPLASMIC.
FT   TRANSMEM    419    438
FT   DOMAIN      439    485       CYTOPLASMIC.
//
ID   CU098_000020
FT   DOMAIN        1    348       NON CYTOPLASMIC.
//