|
4
|
1 #!/usr/bin/env python3
|
|
|
2 ''' configuration file to set up the paths and constants '''
|
|
|
3 import os
|
|
|
4
|
|
|
5 ######## PROFREP #######################################################
|
|
|
6 ## Constansts
|
|
|
7 N_segment = 50
|
|
|
8 MAX_FILES_SUBPROFILES = 1000
|
|
|
9 MAX_PIC_NUM = 50
|
|
|
10 IMAGE_RES = 300
|
|
|
11 FASTA_LINE = 60
|
|
|
12 SEQ_LEN_VIZ = 200000
|
|
|
13 FORBIDDEN_CHARS = "\\/"
|
|
|
14 HTML_STR = '''
|
|
|
15 <!DOCTYPE html>
|
|
|
16 <html>
|
|
|
17 <body>
|
|
|
18 <h2>PROFREP OUTPUT</h2>
|
|
|
19 <h4> Sequences processed: </h4>
|
|
|
20 {}
|
|
|
21 <h4> Total length: </h4>
|
|
|
22 <pre> {} bp </pre>
|
|
|
23 <h4> Database: </h4>
|
|
|
24 <pre> {} </pre>
|
|
|
25 <hr>
|
|
|
26 <h3> Repetitive profile(s)</h3> </br>
|
|
|
27 {} <br/>
|
|
|
28 <h4>References: </h4>
|
|
|
29 {}
|
|
|
30 </h6>
|
|
|
31 </body>
|
|
|
32 </html>
|
|
|
33 '''
|
|
|
34
|
|
|
35 ## IO
|
|
|
36 DOMAINS_GFF = "output_domains.gff"
|
|
|
37 N_GFF = "N_regions.gff"
|
|
|
38 REPEATS_GFF = "output_repeats.gff"
|
|
|
39 HTML = "output.html"
|
|
|
40 LOG_FILE = "log.txt"
|
|
|
41 PROFREP_DATA = "tool_data/profrep"
|
|
|
42 PROFREP_TBL = "prepared_datasets.txt"
|
|
|
43 PROFREP_OUTPUT_DIR = "profrep_output_dir"
|
|
|
44 ## JBrowse and Tracks Conf
|
|
|
45 jbrowse_data_dir = "data"
|
|
|
46 JSON_CONF_R = """{"hooks" : {"modify": "function( track, f, fdiv ) {fdiv.style.backgroundColor = '#278ECF'}"}}"""
|
|
|
47 JSON_CONF_N = """{"hooks" : {"modify": "function( track, f, fdiv ) {fdiv.style.background = '#474747'}"}}"""
|
|
|
48 COLORS_HEX = ["#7F7F7F", "#00FF00", "#0000FF", "#FF0000", "#01FFFE", "#FFA6FE",
|
|
|
49 "#FFDB66", "#006401", "#010067", "#95003A", "#007DB5", "#FF00F6",
|
|
|
50 "#774D00", "#90FB92", "#0076FF", "#D5FF00", "#FF937E", "#6A826C",
|
|
|
51 "#FF029D", "#FE8900", "#7A4782", "#7E2DD2", "#85A900", "#FF0056",
|
|
|
52 "#A42400", "#00AE7E", "#683D3B", "#BDC6FF", "#263400", "#BDD393",
|
|
|
53 "#00B917", "#9E008E", "#001544", "#C28C9F", "#FF74A3", "#01D0FF",
|
|
|
54 "#004754", "#E56FFE", "#788231", "#0E4CA1", "#91D0CB", "#BE9970",
|
|
|
55 "#968AE8", "#BB8800", "#43002C", "#DEFF74", "#00FFC6", "#FFE502",
|
|
|
56 "#620E00", "#008F9C", "#98FF52", "#7544B1", "#B500FF", "#00FF78",
|
|
|
57 "#FF6E41", "#005F39", "#6B6882", "#5FAD4E", "#A75740", "#A5FFD2",
|
|
|
58 "#FFB167", "#009BFF", "#E85EBE"]
|
|
|
59 COLORS_RGB = ["127,127,127", "0,255,0", "0,0,255", "255,0,0", "1,255,254",
|
|
|
60 "255,166,254", "255,219,102", "0,100,1", "1,0,103", "149,0,58",
|
|
|
61 "0,125,181", "255,0,246", "119,77,0", "144,251,146", "0,118,255",
|
|
|
62 "213,255,0", "255,147,126", "106,130,108", "255,2,157",
|
|
|
63 "254,137,0", "122,71,130", "126,45,210", "133,169,0", "255,0,86",
|
|
|
64 "164,36,0", "0,174,126", "104,61,59", "189,198,255", "38,52,0",
|
|
|
65 "189,211,147", "0,185,23", "158,0,142", "0,21,68", "194,140,159",
|
|
|
66 "255,116,163", "1,208,255", "0,71,84", "229,111,254",
|
|
|
67 "120,130,49", "14,76,161", "145,208,203", "190,153,112",
|
|
|
68 "150,138,232", "187,136,0", "67,0,44", "222,255,116",
|
|
|
69 "0,255,198", "255,229,2", "98,14,0", "0,143,156", "152,255,82",
|
|
|
70 "117,68,177", "181,0,255", "0,255,120", "255,110,65", "0,95,57",
|
|
|
71 "107,104,130", "95,173,78", "167,87,64", "165,255,210",
|
|
|
72 "255,177,103", "0,155,255", "232,94,190"]
|
|
|
73 TRACK_LIST = '''
|
|
|
74 \t,{}\n
|
|
|
75 \t"storeClass" : "JBrowse/Store/SeqFeature/BigWig",
|
|
|
76 \t"urlTemplate" : "{}",
|
|
|
77 \t"type" : "JBrowse/View/Track/Wiggle/XYPlot",
|
|
|
78 \t"label" : "{}",
|
|
|
79 \t"key" : "{}",
|
|
|
80 \t"style": {}
|
|
|
81 \t\t"pos_color": "{}"
|
|
|
82 \t {},
|
|
|
83 \t"scale" : "log"
|
|
|
84 \t{}\n
|
|
|
85 '''
|
|
|
86
|
|
|
87 ## GFF tracks
|
|
|
88 HEADER_GFF = "##gff-version 3"
|
|
|
89 SOURCE_PROFREP = "profrep"
|
|
|
90 SOURCE_DANTE = "dante"
|
|
|
91 PHASE = "."
|
|
|
92 DOMAINS_FEATURE = "protein_domain"
|
|
|
93 REPEATS_FEATURE = "repeat"
|
|
|
94 N_NAME = "N"
|
|
|
95 N_FEATURE = "N_region"
|
|
|
96 HEADER_WIG = "variableStep\tchrom="
|
|
|
97 GFF_EMPTY = "."
|
|
|
98
|
|
|
99 ######### BIG WIG ######################################################
|
|
|
100 CHROM_SIZES_FILE = "chrom_sizes.txt"
|
|
|
101
|
|
|
102 ######### EXTRACT_DATA_DOR_PROFREP #####################################
|
|
|
103 HITSORT_CLS = "seqclust/clustering/hitsort.cls"
|
|
|
104 READS_ALL = "seqclust/sequences/sequences.fasta"
|
|
|
105 ANNOTATION = "PROFREP_CLASSIFICATION_TEMPLATE.csv"
|
|
|
106
|
|
|
107 ######### PROFREP_DB_REDUCING ##########################################
|
|
|
108 MEM_LIM = 1500 # MB
|
|
|
109 CLS_REDUCED = "hitsort_reduced.cls"
|
|
|
110 READS_ALL_REDUCED = "reads_all_reduced"
|
|
|
111
|
|
|
112 ######### PROFREP_REFINING #############################################
|
|
|
113 WITH_DOMAINS = "mobile_element"
|
|
|
114 QUALITY_DIFF_TO_REMOVE = 0.05 # 5% tolerance of PID
|
|
|
115
|
|
|
116 ######### DANTE ##############################################
|
|
|
117 MAIN_GIT_DIR = os.path.dirname(os.path.realpath(__file__))
|
|
|
118 DOMAINS_DATA = os.path.join(MAIN_GIT_DIR, "domains_data")
|
|
|
119 TMP = "tmp"
|
|
|
120 SC_MATRIX = os.path.join(DOMAINS_DATA, "blosum80.txt")
|
|
|
121 AMBIGUOUS_TAG = "Ambiguous_domain"
|
|
|
122 ## IO
|
|
|
123 CLASS_FILE = "ALL.classification-new"
|
|
|
124 LAST_DB_FILE = "ALL_protein-domains_05.fasta"
|
|
|
125 DOM_PROT_SEQ = "dom_prot_seq.fa"
|
|
|
126 FILT_DOM_GFF = "domains_filtered.gff"
|
|
|
127 EXTRACT_DOM_STAT = "domains_counts.txt"
|
|
|
128 EXTRACT_OUT_DIR = "extracted_domains"
|
