prepare for the first release

install.sh

@@ -25,8 +25,6 @@ fi
 
 
 
-
-
 # Set default options.
 CLEAN_ONLY=false
 FORCE=false
@@ -43,77 +41,79 @@ fi
 
 
 declare -A DB_ID_HUMAN=(
+['viral_vipr']='1fIxhnwNSPj6NL2R44bqYkYu2T8OLfqpk'
+['fungi']='1yBeBjnrnHtxZliruu3oC8NjZ3wHg-WQ3'
 ['BWAindex']='19Uscw8KrPyUiuPcErrXpyOxN0PqUtbOZ'
+['protozoa']='1_dPn8kk3I--Icy0gwTorFneV1sor1dU2'
 )
 
 declare -A DB_MD5_HUMAN=(
+['viral_vipr']='9dce447328dfbc3a62cc7dd5b052242f'
+['fungi']='9f2d304fd5c49981682b2bb7a900a30e'
 ['BWAindex']='4f009e3732d9f513e7b19b58edc41c13'
+['protozoa']='23e12115a5e9d526553c901e772731f5'
 )
 
 
 ORGANISM='human'
-download_list=$'BWAindex'
+download_list=$'BWAindex\nviral_vipr\nfungi\nprotozoa'
+
+
+
+
+
+
 
 cd "$DB_LOCATION"
 mkdir "db_$ORGANISM"
 cd "db_$ORGANISM"
 
 
-#echo '----- Downloading hg38 human reference ----------------------------------------------------'
-#for download in $download_list; do
-#echo "Downloading item: $download for $ORGANISM"
-#success=false
-#while [ $success = false ]; do
-#case "$ORGANISM" in
-#human)
-#db_id="${DB_ID_HUMAN[$download]}"
-#db_md5="${DB_MD5_HUMAN[$download]}"
-#;;
-#mouse)
-#db_id="${DB_ID_MOUSE[$download]}"
-#db_md5="${DB_MD5_MOUSE[$download]}"
-#;;
-#*)
-#echo 'Error: Unknown ORGANISM.' >&2
-#exit 1
-#;;
-#esac
-#confirm_code=`curl --silent --insecure --cookie-jar cookies.txt "https://docs.google.com/uc?export=download&id=$db_id" | sed -rn 's .*confirm=([0-9A-Za-z_]+).* \1\n p'`
 
-#curl --location --insecure --cookie cookies.txt \"https://docs.google.com/uc?export=download&confirm=$confirm_code&id=$db_id\"  >$download.tar.gz
 
 
+echo '----- Downloading human and mirobial references ----------------------------------------------------'
+for download in $download_list; do
+echo "Downloading item: $download for $ORGANISM"
+success=false
+while [ $success = false ]; do
+case "$ORGANISM" in
+human)
+db_id="${DB_ID_HUMAN[$download]}"
+db_md5="${DB_MD5_HUMAN[$download]}"
+;;
+mouse)
+db_id="${DB_ID_MOUSE[$download]}"
+db_md5="${DB_MD5_MOUSE[$download]}"
+;;
+*)
+echo 'Error: Unknown ORGANISM.' >&2
+exit 1
+;;
+esac
+confirm_code=`curl --silent --insecure --cookie-jar cookies.txt "https://docs.google.com/uc?export=download&id=$db_id" | sed -rn 's .*confirm=([0-9A-Za-z_]+).* \1\n p'`
+curl --location --insecure --cookie cookies.txt "https://docs.google.com/uc?export=download&confirm=$confirm_code&id=$db_id"  >$download.tar.gz
+
 
 
 
-#rm cookies.txt
-#if [ `md5sum "$download.tar.gz" | sed 's \(.*\)\ .* \1 '` = "$db_md5" ]; then
-#tar -zxvf "$download.tar.gz"
-#rm "$download.tar.gz"
-#success=true
-#else
-#echo "Download of $download for $ORGANISM failed (checksum" \
-#'mismatch. Retrying.'
-#fi
-#done
-#done
 
+rm cookies.txt
+if [ `md5sum "$download.tar.gz" | sed 's \(.*\)\ .* \1 '` = "$db_md5" ]; then
+tar -zxvf "$download.tar.gz"
+rm "$download.tar.gz"
+success=true
+else
+echo "Download of $download for $ORGANISM failed (checksum" \
+'mismatch. Retrying.'
+fi
+done
+done
 
 
 
-#ADD----------
-#./conda install -c dranew hmmcopy_utils
-#git clone https://github.com/broadinstitute/ichorCNA.git
-#conda install -c r r-base
-#old conda install -c bioconda r-optparse
-#old source("https://bioconductor.org/biocLite.R")
-#old biocLite("HMMcopy")
 
-#biocLite("GenomeInfoDb")
 
-#install.packages("devtools")
-#library(devtools)
-#install_github("broadinstitute/ichorCNA", "--no-docs")
 
 # ------------------------------------------------------------------------------
 # DOWNLOAD TOOLS
@@ -143,10 +143,10 @@ else
 
     #Download megahit
     echo '----- Downloading Megahit --------------------------------------------------'
-    #git clone https://github.com/voutcn/megahit.git
-    #cd megahit
-    #make
-    #cd ..
+    git clone https://github.com/voutcn/megahit.git
+    cd megahit
+    make
+    cd ..
 
 
 

needle.sh

@@ -68,9 +68,9 @@ UNMAPPED=${SAMPLE}.unmapped.fastq
 
 
 
-bwa mem -a ${DIR_CODE}/db.microbiome/viral.vipr/NONFLU_All.fastq $UNMAPPED | samtools view -S -b -F 4 - | samtools sort - >${SAMPLE}.virus.bam
-bwa mem -a ${DIR_CODE}/db.microbiome/fungi/fungi.ncbi.february.3.2018.fasta $UNMAPPED | samtools view -S -b -F 4 - |  samtools sort - >${SAMPLE}.fungi.bam
-bwa mem -a ${DIR_CODE}/db.microbiome/protozoa/protozoa.ncbi.february.3.2018.fasta $UNMAPPED | samtools view -S -b -F 4 - | samtools sort - >${SAMPLE}.protozoa.bam
+bwa mem -a ${DB}/viral.vipr/NONFLU_All.fastq $UNMAPPED | samtools view -S -b -F 4 - | samtools sort - >${SAMPLE}.virus.bam
+bwa mem -a ${DB}/fungi/fungi.ncbi.february.3.2018.fasta $UNMAPPED | samtools view -S -b -F 4 - |  samtools sort - >${SAMPLE}.fungi.bam
+bwa mem -a ${DB}/protozoa/protozoa.ncbi.february.3.2018.fasta $UNMAPPED | samtools view -S -b -F 4 - | samtools sort - >${SAMPLE}.protozoa.bam
 
 samtools index ${SAMPLE}.virus.bam
 samtools index ${SAMPLE}.fungi.bam
@@ -154,9 +154,9 @@ samtools view -F 4  ${SAMPLE}.megahit.contigs.protozoa.bam | grep -v -e 'XA:Z:'
 echo "-----------------------------------------------------"
 echo "Map assembled contigs onto the microbial references"
 
-bwa mem -a ${DIR_CODE}/db.microbiome/viral.vipr/NONFLU_All.fastq ${SAMPLE}.virus.megahit.contigs.fa | samtools view -bS -F 4 -  >${SAMPLE}.virus.megahit.contigs.SV.bam
-bwa mem -a ${DIR_CODE}/db.microbiome/fungi/fungi.ncbi.february.3.2018.fasta ${SAMPLE}.fungi.megahit.contigs.fa  | samtools view -bS -F 4 -  >${SAMPLE}.fungi.megahit.contigs.SV.bam
-bwa mem -a ${DIR_CODE}/db.microbiome/protozoa/protozoa.ncbi.february.3.2018.fasta ${SAMPLE}.protozoa.megahit.contigs.fa  | samtools view -bS -F 4 ->${SAMPLE}.protozoa.megahit.contigs.SV.bam
+bwa mem -a ${DB}/viral.vipr/NONFLU_All.fastq ${SAMPLE}.virus.megahit.contigs.fa | samtools view -bS -F 4 -  >${SAMPLE}.virus.megahit.contigs.SV.bam
+bwa mem -a ${DB}/fungi/fungi.ncbi.february.3.2018.fasta ${SAMPLE}.fungi.megahit.contigs.fa  | samtools view -bS -F 4 -  >${SAMPLE}.fungi.megahit.contigs.SV.bam
+bwa mem -a ${DB}/protozoa/protozoa.ncbi.february.3.2018.fasta ${SAMPLE}.protozoa.megahit.contigs.fa  | samtools view -bS -F 4 ->${SAMPLE}.protozoa.megahit.contigs.SV.bam