conf.py 3.75 KB
Newer Older
aknecht2's avatar
aknecht2 committed
1
2
import string

3
4
5
6
7
8
# Module directory
job_modules = "jobs/modules/"

# Job params directory
job_params = "jobs/params/"

9
10
11
# Job wrappers directory
job_wrappers = "jobs/wrappers/"

12
13
14
# Job scripts directory
job_scripts = "jobs/scripts/"

15
16
17
18
19
20
21
22
23
24
25
26
# SYSTEM PATH, probably /bin
system_path = "/bin/"

# System commands mv, cp, sort, zcat, awk e.t.c
system_commands = [
    "mv",
    "cp",
    "sort",
    "zcat",
    "awk"
]

27
28
29
30
31
32
# Current align tools
align_tools = [
    "bwa",
    "bowtie2"
]

33
# Current peak calling tools
34
35
peak_tools = [
    "spp",
36
    "macs2",
37
    "gem",
38
    "peakranger",
aknecht2's avatar
aknecht2 committed
39
40
    "ccat",
    "music"
41
42
]

43
44
45
46
executables = [
    "bedtools",
    "samtools",
    "idr",
47
48
49
50
51
52
53
    "picard",
    "bwa",
    "bowtie2",
    "macs2",
    "gem",
    "peakranger",
    "MUSIC",
54
55
    "CCAT",
    "run_spp_nodups"
56
]
57

58
59
60
61
62
63
# Peak_type validation
peak_types = {
    "spp": ["narrow", "broad"],
    "macs2": ["narrow", "broad"],
    "gem": ["narrow"],
    "peakranger": ["narrow"],
aknecht2's avatar
aknecht2 committed
64
65
    "ccat": ["broad"],
    "music": ["narrow", "punctate", "broad"]
66
67
}

68
69
# File extensions
file_extensions = {
70
    "genome_index": ["fa", "fna"],
71
    "fastq": ["fastq", "fastq.gz"],
72
    "sai": ["sai"],
73
74
    "sam": ["sam"],
    "bam": ["bam"],
75
    "bed": ["bed", "peak", "region", "narrowPeak", "broadPeak", "tagAlign", "narrowPeak.gz", "broadPeak.gz"],
aknecht2's avatar
aknecht2 committed
76
    "bin": ["bin"],
77
    "bwa_genome": ["amb", "ann", "bwt", "pac", "sa"],
78
    "bowtie2_genome": ["1.bt2", "2.bt2", "3.bt2", "4.bt2", "rev.1.bt2", "rev.2.bt2"],
aknecht2's avatar
aknecht2 committed
79
    "quality": ["quality"],
80
81
    "qc": ["qc"],
    "pdf": ["pdf"],
82
    "ccscore": ["ccscore"],
83
84
    "xls": ["xls"],
    "yaml": ["yaml"],
Adam Caprez's avatar
Adam Caprez committed
85
    "result": ["bed", "narrowPeak", "broadPeak", "tagAlign", "bam"],
86
    "txt": ["txt"],
87
    "chrom_sizes": ["sizes"],
88
    "read_dist" : ["txt"],
89
90
91
    "plaintext": ["txt", "md"],
    "ccat_conf": ["txt", "conf"],
    "log": ["log"]
92
}
aknecht2's avatar
aknecht2 committed
93
file_extensions["any"] = list(string.ascii_lowercase)
94

Adam Caprez's avatar
Adam Caprez committed
95
96
# list of resources that can be specified per job (step) in
# the workflow and corresponding Pegasus profile info
97
98
99
100
101
102
103
104
resources = {
    "walltime": {
        "namespace": "globus",
        "key": "maxwalltime"
    },
    "memory": {
        "namespace": "condor",
        "key": "request_memory"
105
106
    },
    "cores": {
aknecht2's avatar
aknecht2 committed
107
        "namespace": "pegasus",
108
        "key": "cores"
109
110
111
112
    },
    "nodes": {
        "namespace": "pegasus",
        "key": "nodes"
113
    }
Adam Caprez's avatar
Adam Caprez committed
114
115
}

116
117
# Defines the types of input / output arguments
# argument -> Any non file argument
118
119
#   string -> Any string argument
#   numeric -> Numeric arguments.
120
# file -> Any file argument that isn't redirected
121
122
123
#   file -> Normal file arguments.
#   stdout -> Any file argument that is redirected from stdout
#   stderr -> Any file argument that is redirected from stderr
124
125
# list -> For variable argument inputs
#   list -> Currently a list of files.
126
argument_types = {
127
    "argument": ["string", "numeric"],
128
    "file": ["file", "rawfile", "stdout", "stderr"],
aknecht2's avatar
aknecht2 committed
129
130
    "list": ["list"],
    "folder": ["folder", "rawfolder"]
131
132
133
134
135
}

# Defines information about arguments
argument_keys = {
    "required": ["type", "changeable", "has_value"],
136
    "optional": ["required", "default", "file_type", "path"]
137
138
139
140
141
142
}

# workflow_job keys
job_keys = {
    "required": ["inputs", "additional_inputs", "outputs", "command", "arguments"] + resources.keys(),
    "optional": []
143
}
144

145
146
# param keys
param_keys = {
147
148
    "required": [],
    "optional": ["arguments"] + resources.keys()
149
150
}

151
152
153
154
155
file_list_keys = {
    "required": ["name", "type"],
    "optional": ["file_type"]
}

156
157
# workflow order
workflow = ["align", "remove_duplicates", "peak_calling"]
158
159
160
161
162
163
164
165
166
167
168
169

# genome info
genomes = {
    "bwa": {
        "base_file": file_extensions["genome_index"],
        "additional_files": file_extensions["bwa_genome"]
    },
    "bowtie2": {
        "base_file": file_extensions["genome_index"],
        "additional_files": file_extensions["bowtie2_genome"]
    }
}