#P. citri all data (VF+MF susphire and Edinburgh all Pcitri samples) RNAspades v3.13.0 transcriptome assembly
#run on bufo

cd /DATA/markop/Pcitri_rnaspades_alldata


#InterProScan (IPS) all soft-filtered contigs from spades

mkdir transeq 

#translate nt -> aa
/DATA/markop/my_interproscan/interproscan-5.32-71.0/bin/nucleotide/translate -i ./output/soft_filtered_transcripts.fasta -o ./transeq/soft_filtered_transcripts.pep

#filter to longest 3 orfs per contig
python3 /DATA/markop/my_interproscan/interproscan-5.32-71.0/bin/nucleotide/parseOrfs.py 3 ./transeq/soft_filtered_transcripts.pep ./transeq/soft_filtered_transcripts.pep.filtered

#IPS run on translated aa seqs without precalc queries
mkdir IPS

/DATA/markop/my_interproscan/interproscan-5.32-71.0/interproscan.sh \
--appl CDD,HAMAP,PANTHER,Pfam,PIRSF,PRINTS,PROSITEPROFILES,SMART,TIGRFAM,PROSITEPATTERNS,Gene3D,SFLD,SUPERFAMILY \
--seqtype p \
-i ./transeq/soft_filtered_transcripts.pep.filtered \
-iprlookup \
-goterms \
--output-dir IPS \
--disable-precalc \
--formats TSV