global_config_bash.rc

#!/bin/bash

#################################################
#                                               #
# Default config file for ecDNA IMPACT analysis #
#                                               #
#################################################

projectName="BB_EchoCaller_v2_50Pediatric"

###############
# directories #
###############

# Directory containing input folder
dataDirectory="../data"

# Directory containing inputs
inputDirectory="${dataDirectory}/input"

# Manifest directory
manifestDirectory="${inputDirectory}/manifest/${projectName}"

# Output, log and flag directories
outputDirectory="${dataDirectory}/output/output_${projectName}"
logDirectory="${dataDirectory}/log/log_${projectName}"
flagDirectory="${dataDirectory}/flag/flag_${projectName}"

# For the initial echo calls
echoOutputDirectory="${outputDirectory}/echoCalls"
echoLogDirectory="${logDirectory}/echoCalls"
echoFlagDirectory="${flagDirectory}/echoCalls"

# For the merged files
mergedOutputDirectory="${outputDirectory}/merged"

# For facets sample calls
sampleFacetsDirectory="${outputDirectory}/sampleFacets"

# For facets gene calls
facetsOutputDirectory="${outputDirectory}/facetsCalls"
facetsLogDirectory="${logDirectory}/facetsCalls"
facetsFlagDirectory="${flagDirectory}/facetsCalls"

###########
# samples #
###########

# Subset of samples to use
sampleFull="${manifestDirectory}/FileB.tsv"

# Files that will be created
sampleInfoFull="${manifestDirectory}/FileA.tsv"
sampleSubset="${manifestDirectory}/FileB_run.tsv"
sampleInfoSubset="${manifestDirectory}/FileA_run.tsv"

# Columns of information in the sampleTracker (0 indexed)
sampleIDColumn=0
tumorPurityColumn=1
somaticStatusColumn=2

# True if we want to use facets purity instead of pathology (True or False)
useFacetsPurity=False
defaultPurity=20

# Analysis type (1 or 2)
aType=1
seqType=IMPACT

###############
# clusterInfo #
###############

# Number of CPUs
clusterCPUNum=4

# Memory per CPU
clusterMemory=64

# Allotted time (format: 72:00 -> 72 hours or 72 -> 72 hours)
clusterTime="72"

####################################
# Access to cBioPortal Private API #
####################################

dataAccessToken="/home/yuk3/cbioportal_data_access_token.txt"

###################
# Reference Files #
###################

keyFile="/juno/dmp/request/12-245/key.txt"
singularity_cache=/home/sumans/.singularity/cache

image_echoPreProcessor="boundlessbio-echo-preprocessor-v2.0.4.img"
image_echoCaller="boundlessbio-echo-caller-v2.4.0.img"
imagePath_echoPreProcessor=${singularity_cache}/${image_echoPreProcessor}
imagePath_echoCaller=${singularity_cache}/${image_echoCaller}

inputDirectory=$(readlink -f "$inputDirectory")
bedFolder=${inputDirectory}/beds
ANNOTATION_FILE=${inputDirectory}/references/refFlat_withoutPrefix.txt
EXCLUDE_FILE=${inputDirectory}/references/human.hg19.excl.tsv

mafFile="data_mutations_extended.txt"
mafPath=/work/access/production/resources/cbioportal/current/msk_solid_heme

GENOME_VERSION=hg19

refFile1="b37.fasta"
refFile2="GRCh37_plus_virus.fa"