#Centrifuge taxonomic classification of potato cultivar specific transcriptomes

cd _A_08_centrifuge_3cvs-GFFmerged

conda activate centrifuge


#run centrifuge version 1.0.4 and generate kraken report-files for use in pavian
#potato taxid is 4113 (used for --host-taxid parameter)

centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ../../../_S_03_stCuSTr/_A_03.2_components/output/fasta_1A_rep-per-cv/stCuSTr-P_tr_representatives.fasta -S ./output/PW363_centrifuge.txt --report-file ./output/PW363_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/PW363_centrifuge.txt > ./output/PW363_kreport.txt

centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ../../../_S_03_stCuSTr/_A_03.2_components/output/fasta_1A_rep-per-cv/stCuSTr-D_tr_representatives.fasta -S ./output/Desiree_centrifuge.txt --report-file ./output/Desiree_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/Desiree_centrifuge.txt > ./output/Desiree_kreport.txt

centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ../../../_S_03_stCuSTr/_A_03.2_components/output/fasta_1A_rep-per-cv/stCuSTr-R_tr_representatives.fasta -S ./output/Rywal_centrifuge.txt --report-file ./output/Rywal_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/Rywal_centrifuge.txt > ./output/Rywal_kreport.txt


#run centrifuge for singleton transcripts only, same parameters as above

centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ./intermediate/trueSingle.DM.cds.fasta -S ./output/trueSingle.DM.cds_centrifuge.txt --report-file ./output/trueSingle.DM.cds_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/trueSingle.DM.cds_centrifuge.txt > ./output/trueSingle.DM.cds_kreport.txt

centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ./intermediate/trueSingle.polyploid.cds.fasta -S ./output/trueSingle.polyploid.cds_centrifuge.txt --report-file ./output/trueSingle.polyploid.cds_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/trueSingle.polyploid.cds_centrifuge.txt > ./output/trueSingle.polyploid.cds_kreport.txt


centrifuge -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt -f -U ./intermediate/trueSingle.polyploid.tr.fasta -S ./output/trueSingle.polyploid.tr_centrifuge.txt --report-file ./output/trueSingle.polyploid.tr_report.txt --host-taxid 4113 --seed 42 -p 30
centrifuge-kreport -x /NFSi-bioDBs/centrifuge/dbs_v2018/nt ./output/trueSingle.polyploid.tr_centrifuge.txt > ./output/trueSingle.polyploid.tr_kreport.txt

#RUN PAVIAN AND GENERATE SANKEY PLOTS from kreport files!
https://fbreitwieser.shinyapps.io/pavian/