#centrifuge on ibis2
# run on 1.2.2018
# nt db from Nov 27  2016 

###################################
###################################
### shotgun DNA-seq CPB samples ###
###################################
###################################

cd /FITO_ws/markop/beetle_metatranscriptomics/CPB_gut_metagenome_DNA-seq/

source /usr/local/centrifuge/setenv.sh

## round one of sequencing (raw data here: http://stork/dirindex/fitostorage/datarepo/ngs/NGS_mpe_2018_CPB-gut-metagenome_DNAseq/original_data_pt1/)

#zymo standard
centrifuge -x /DATA/centrifuge_data/nt -1  RI0000_1.fq -2 RI0000_2.fq -S RI000_centrifuge --report-file RI0000_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI000_centrifuge > RI000_centrifuge_kreport.txt

#sample 1
centrifuge -x /DATA/centrifuge_data/nt -1  RI0325_1.fq -2 RI0325_2.fq -S RI0325_centrifuge --report-file RI0325_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0325_centrifuge > RI0325_centrifuge_kreport.txt

#sample 2
centrifuge -x /DATA/centrifuge_data/nt -1  RI0328_1.fq -2 RI0328_2.fq -S RI0328_centrifuge --report-file RI0328_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0328_centrifuge > RI0328_centrifuge_kreport.txt

#sample 3
centrifuge -x /DATA/centrifuge_data/nt -1  RI0331_1.fq -2 RI0331_2.fq -S RI0331_centrifuge --report-file RI0331_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0331_centrifuge > RI0331_centrifuge_kreport.txt

#sample 4
centrifuge -x /DATA/centrifuge_data/nt -1  RI0332_1.fq -2 RI0332_2.fq -S RI0332_centrifuge --report-file RI0332_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0332_centrifuge > RI0332_centrifuge_kreport.txt
#done


## round two of sequencing (raw data here: http://stork/dirindex/fitostorage/datarepo/ngs/NGS_mpe_2018_CPB-gut-metagenome_DNAseq/original_data_pt2/)

#sample 5
centrifuge -x /DATA/centrifuge_data/nt -1  RI0326_1.fq -2 RI0326_2.fq -S RI0326_centrifuge --report-file RI0326_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0326_centrifuge > RI0326_centrifuge_kreport.txt

#sample 6
centrifuge -x /DATA/centrifuge_data/nt -1  RI0327_1.fq -2 RI0327_2.fq -S RI0327_centrifuge --report-file RI0327_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0327_centrifuge > RI0327_centrifuge_kreport.txt

#sample 7
centrifuge -x /DATA/centrifuge_data/nt -1  RI0329_1.fq -2 RI0329_2.fq -S RI0329_centrifuge --report-file RI0329_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0329_centrifuge > RI0329_centrifuge_kreport.txt

#sample 8
centrifuge -x /DATA/centrifuge_data/nt -1  RI0330_1.fq -2 RI0330_2.fq -S RI0330_centrifuge --report-file RI0330_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0330_centrifuge > RI0330_centrifuge_kreport.txt

#sample 9
centrifuge -x /DATA/centrifuge_data/nt -1  RI0333_1.fq -2 RI0333_2.fq -S RI0333_centrifuge --report-file RI0333_report.txt
centrifuge-kreport -x /DATA/centrifuge_data/nt RI0333_centrifuge > RI0333_centrifuge_kreport.txt

## convert to metaphlan-style files
cd /FITO_ws/markop/beetle_metatranscriptomics/CPB_gut_metagenome_DNA-seq
Bracken/src/kreport2mpa.py -r RI0325_centrifuge_kreport.txt -o RI0325_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0326_centrifuge_kreport.txt -o RI0326_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0327_centrifuge_kreport.txt -o RI0327_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0328_centrifuge_kreport.txt -o RI0328_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0329_centrifuge_kreport.txt -o RI0329_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0330_centrifuge_kreport.txt -o RI0330_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0331_centrifuge_kreport.txt -o RI0331_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0332_centrifuge_kreport.txt -o RI0332_centrifuge_kreport_mpa.txt --intermediate-ranks
Bracken/src/kreport2mpa.py -r RI0333_centrifuge_kreport.txt -o RI0333_centrifuge_kreport_mpa.txt --intermediate-ranks

#merge samples for import (R phyloseq library)
python merge_metaphlan_tables.py RI000_centrifuge_kreport_mpa.txt RI0325_centrifuge_kreport_mpa.txt RI0326_centrifuge_kreport_mpa.txt RI0327_centrifuge_kreport_mpa.txt RI0328_centrifuge_kreport_mpa.txt RI0329_centrifuge_kreport_mpa.txt RI0330_centrifuge_kreport_mpa.txt RI0331_centrifuge_kreport_mpa.txt RI0332_centrifuge_kreport_mpa.txt RI0333_centrifuge_kreport_mpa.txt > DNA_seq_combined_mpa.txt
cut -f2- DNA_seq_combined_mpa.txt > DNA_seq_combined_mpa_OTU.txt
cut -f1 DNA_seq_combined_mpa.txt > DNA_seq_combined_mpa_TAX.txt
# use this two files to import into R -- see script file metagenomics_plots.R