intSiteCallerConfig.yaml

# Log file path
logFile : intSiteCaller.log

# Debug
debug : No
forceQsubPath : No

# Run id
runId : testRun


# Parallelize the calculations with bsub or qsub
parallelize : Yes


# Parallelization scheduler. If bsub is set to No the qsub will be used by default.
# This option will be ignored in parallelize is set to 'No'.
bsub : No


# Maximum size for each sequence file chunk
chunkSize : 30000


# Log in to system hosting vector and sequencing data files
remoteUser : everett@microb120.med.upenn.edu


# Directory that holds the vector information file defiled in the vectorSeq column of the provided sampleInfo.csv file  
vectorDataPath : ../intSiteValidation


#  Directory that holds the R1, R2, and I1 sequencing run gzipped FASTQ files 
SequencingFiles:
  I1 :  ../intSiteValidation/Data/Undetermined_S0_L001_I1_001.fastq.gz
  R1 :  ../intSiteValidation/Data/Undetermined_S0_L001_R1_001.fastq.gz
  R2 :  ../intSiteValidation/Data/Undetermined_S0_L001_R2_001.fastq.gz


# Database connection group defined in ~/.my.cnf
databaseConnectionGroup :  intsites_miseq.read


# Processing parameters
ProcessingParameters:
  qualityThreshold     : '?'
  badQualityBases      : 5
  qualitySlidingWindow : 10
  mingDNA              : 20
  minPctIdent          : 95
  maxAlignStart        : 5
  maxFragLength        : 2500   
  refGenome            : hg18