########################################
						### recentrifuge CPB DNA-seq samples ###
						########################################

#run on bufo_Mint

conda activate recentrifuge

cd /DATA/markop
mkdir CPB_rcf_DNAseq

cd CPB_rcf_DNAseq

#made a softlink "input" for "/NFSd-FITO_ws/markop/beetle_metatranscriptomics/_fromBUFO2/CPB_gut_metagenome_DNA-seq/"

#retaxdump
#downloaded https://drive.google.com/uc?id=1UqDo_ggzKzD35ZHkQGUldd__jVmbHI_R&export=download

# RI0000 used as control sample although it's not a blank control
#excluding taxa assigned to viridiplantae (taxid 33090), metazoa = multicellular animals (taxid 33208), Phix174microvirus (taxid 1910954), unclassified sequences (taxid 12908) and other sequences (taxid 28384).
rcf \
-n ./taxdmp \
-c 1 \
-f ./input/RI0000_centrifuge \
-f ./input/RI0325_centrifuge \
-f ./input/RI0326_centrifuge \
-f ./input/RI0327_centrifuge \
-f ./input/RI0328_centrifuge \
-f ./input/RI0329_centrifuge \
-f ./input/RI0330_centrifuge \
-f ./input/RI0331_centrifuge \
-f ./input/RI0332_centrifuge \
-f ./input/RI0333_centrifuge \
-x 33090 -x 33208 -x 1910954 -x 12908 -x 28384 \
-o CPB-DNAseq_RI0000asCTRL.rcf.html

# no control!
#excluding taxa assigned to viridiplantae (taxid 33090), metazoa = multicellular animals (taxid 33208), Phix174microvirus (taxid 1910954), unclassified sequences (taxid 12908) and other sequences (taxid 28384).
rcf \
-n ./taxdmp \
-c 0 \
-f ./input/RI0325_centrifuge \
-f ./input/RI0326_centrifuge \
-f ./input/RI0327_centrifuge \
-f ./input/RI0328_centrifuge \
-f ./input/RI0329_centrifuge \
-f ./input/RI0330_centrifuge \
-f ./input/RI0331_centrifuge \
-f ./input/RI0332_centrifuge \
-f ./input/RI0333_centrifuge \
-x 33090 -x 33208 -x 1910954 -x 12908 -x 28384 \
-o CPB-DNAseq_noCTRL.rcf.html



###############
### RCF cut-off to retain only high-scored taxa (minscore 75 as in the Recentrifuge publication Fig S7)

# RI0000 used as control sample although it's not a blank control
#excluding taxa assigned to viridiplantae (taxid 33090), metazoa = multicellular animals (taxid 33208), Phix174microvirus (taxid 1910954), unclassified sequences (taxid 12908) and other sequences (taxid 28384).
rcf \
-n ./taxdmp \
-c 1 \
-f ./input/RI0000_centrifuge \
-f ./input/RI0325_centrifuge \
-f ./input/RI0326_centrifuge \
-f ./input/RI0327_centrifuge \
-f ./input/RI0328_centrifuge \
-f ./input/RI0329_centrifuge \
-f ./input/RI0330_centrifuge \
-f ./input/RI0331_centrifuge \
-f ./input/RI0332_centrifuge \
-f ./input/RI0333_centrifuge \
-x 33090 -x 33208 -x 1910954 -x 12908 -x 28384 \
--minscore 75 \
-o CPB-DNAseq_RI0000asCTRL_cut75.rcf.html

# no control!
#excluding taxa assigned to viridiplantae (taxid 33090), metazoa = multicellular animals (taxid 33208), Phix174microvirus (taxid 1910954), unclassified sequences (taxid 12908) and other sequences (taxid 28384).
rcf \
-n ./taxdmp \
-c 0 \
-f ./input/RI0325_centrifuge \
-f ./input/RI0326_centrifuge \
-f ./input/RI0327_centrifuge \
-f ./input/RI0328_centrifuge \
-f ./input/RI0329_centrifuge \
-f ./input/RI0330_centrifuge \
-f ./input/RI0331_centrifuge \
-f ./input/RI0332_centrifuge \
-f ./input/RI0333_centrifuge \
-x 33090 -x 33208 -x 1910954 -x 12908 -x 28384 \
--minscore 75 \
-o CPB-DNAseq_noCTRL_cut75.rcf.html




cp ./input/*.fq ./fq/
rm ./fq/RI0000*.fq


####################
# extract Serratia marcensens (taxid 615)-assigned reads for de-novo assembly of it's genome
####################

rextract \
-n ./taxdmp \
-f ./input/RI0325_centrifuge \
-i 615 \
-1 ./fq/RI0325_1.fq \
-2 ./fq/RI0325_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0326_centrifuge \
-i 615 \
-1 ./fq/RI0326_1.fq \
-2 ./fq/RI0326_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0327_centrifuge \
-i 615 \
-1 ./fq/RI0327_1.fq \
-2 ./fq/RI0327_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0328_centrifuge \
-i 615 \
-1 ./fq/RI0328_1.fq \
-2 ./fq/RI0328_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0329_centrifuge \
-i 615 \
-1 ./fq/RI0329_1.fq \
-2 ./fq/RI0329_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0330_centrifuge \
-i 615 \
-1 ./fq/RI0330_1.fq \
-2 ./fq/RI0330_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0331_centrifuge \
-i 615 \
-1 ./fq/RI0331_1.fq \
-2 ./fq/RI0331_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0332_centrifuge \
-i 615 \
-1 ./fq/RI0332_1.fq \
-2 ./fq/RI0332_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0333_centrifuge \
-i 615 \
-1 ./fq/RI0333_1.fq \
-2 ./fq/RI0333_2.fq


####################
# extract Lactococcus lactis (taxid 1358)-assigned reads for de-novo assembly of it's genome
####################
rextract \
-n ./taxdmp \
-f ./input/RI0325_centrifuge \
-i 1358 \
-1 ./fq/RI0325_1.fq \
-2 ./fq/RI0325_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0326_centrifuge \
-i 1358 \
-1 ./fq/RI0326_1.fq \
-2 ./fq/RI0326_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0327_centrifuge \
-i 1358 \
-1 ./fq/RI0327_1.fq \
-2 ./fq/RI0327_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0328_centrifuge \
-i 1358 \
-1 ./fq/RI0328_1.fq \
-2 ./fq/RI0328_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0329_centrifuge \
-i 1358 \
-1 ./fq/RI0329_1.fq \
-2 ./fq/RI0329_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0330_centrifuge \
-i 1358 \
-1 ./fq/RI0330_1.fq \
-2 ./fq/RI0330_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0331_centrifuge \
-i 1358 \
-1 ./fq/RI0331_1.fq \
-2 ./fq/RI0331_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0332_centrifuge \
-i 1358 \
-1 ./fq/RI0332_1.fq \
-2 ./fq/RI0332_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0333_centrifuge \
-i 1358 \
-1 ./fq/RI0333_1.fq \
-2 ./fq/RI0333_2.fq

####################
# extract Enterococcus gallinarum (taxid 1353)-assigned reads for de-novo assembly of it's genome
####################
rextract \
-n ./taxdmp \
-f ./input/RI0325_centrifuge \
-i 1353 \
-1 ./fq/RI0325_1.fq \
-2 ./fq/RI0325_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0326_centrifuge \
-i 1353 \
-1 ./fq/RI0326_1.fq \
-2 ./fq/RI0326_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0327_centrifuge \
-i 1353 \
-1 ./fq/RI0327_1.fq \
-2 ./fq/RI0327_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0328_centrifuge \
-i 1353 \
-1 ./fq/RI0328_1.fq \
-2 ./fq/RI0328_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0329_centrifuge \
-i 1353 \
-1 ./fq/RI0329_1.fq \
-2 ./fq/RI0329_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0330_centrifuge \
-i 1353 \
-1 ./fq/RI0330_1.fq \
-2 ./fq/RI0330_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0331_centrifuge \
-i 1353 \
-1 ./fq/RI0331_1.fq \
-2 ./fq/RI0331_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0332_centrifuge \
-i 1353 \
-1 ./fq/RI0332_1.fq \
-2 ./fq/RI0332_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0333_centrifuge \
-i 1353 \
-1 ./fq/RI0333_1.fq \
-2 ./fq/RI0333_2.fq

####################
# extract Enterobacter cloacae (taxid 550)-assigned reads for de-novo assembly of it's genome
####################
rextract \
-n ./taxdmp \
-f ./input/RI0325_centrifuge \
-i 550 \
-1 ./fq/RI0325_1.fq \
-2 ./fq/RI0325_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0326_centrifuge \
-i 550 \
-1 ./fq/RI0326_1.fq \
-2 ./fq/RI0326_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0327_centrifuge \
-i 550 \
-1 ./fq/RI0327_1.fq \
-2 ./fq/RI0327_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0328_centrifuge \
-i 550 \
-1 ./fq/RI0328_1.fq \
-2 ./fq/RI0328_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0329_centrifuge \
-i 550 \
-1 ./fq/RI0329_1.fq \
-2 ./fq/RI0329_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0330_centrifuge \
-i 550 \
-1 ./fq/RI0330_1.fq \
-2 ./fq/RI0330_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0331_centrifuge \
-i 550 \
-1 ./fq/RI0331_1.fq \
-2 ./fq/RI0331_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0332_centrifuge \
-i 550 \
-1 ./fq/RI0332_1.fq \
-2 ./fq/RI0332_2.fq

rextract \
-n ./taxdmp \
-f ./input/RI0333_centrifuge \
-i 550 \
-1 ./fq/RI0333_1.fq \
-2 ./fq/RI0333_2.fq


## delete copied *.fq files and gzip all extracted *.fastq files

rm ./fq/*.fq
pigz -p 32 ./fq/*.fastq