#PBS -N testing_multiple_job_submission
#PBS -l nodes=1:ppn=1
#PBS -l mem=10gb
#PBS -l walltime=6:00:00
#PBS -q iw-shared-6
#PBS -j oe
#PBS -o $HOME/job_output_files/out.chimera_detect.${PBS_JOBID}
#PBS -m e
#PBS -M waoverholt@gmail.com

#dereplicate files
usearch1.7.0 -derep_fulllength ${INFILE} -output ${OUTFILE} -uc ${UCFILE} -sizeout &&

#identify chimeras
usearch1.7.0 -uchime_denovo ${OUTFILE} -nonchimeras ${NONCHIMFILE} &&

usearch1.7.0 -uchime_ref ${NONCHIMFILE} -db /nv/hp10/woverholt3/data/program_files/Silva_ref_dbs/silva.gold.notalign.fasta -nonchimeras ${NONCHIMNONREF} -strand plus &&

#convert uc file to a qiime map file
$HOME/overholt_scripts/convert_uc2map.py ${UCFILE} > ${MAPFILE} &&

#grab all singletons that are not present in the OTU map file
perl -ne 'if ($_ =~ m/>/ && $_ =~ m/size=1;/) { ($ID = $_) =~ s/>(.*);size.*/$1/; print $ID }' ${NONCHIMNONREF} > ${GOODSEQIDS} &&

#good to here: goodseqids = 35436 (all the size one)

#identify chimeric dereps
perl -ne 'if ($_ =~ m/>/ && $_ !~ m/size=1;/) {($ID = $_) =~ s/>(.*);size.*/$1/; print $ID;}' ${NONCHIMNONREF} > ${GOODOTUFILE} &&

#grab all seqids associated with chimeric dereps & add to previously id'd chimeras
join -j 1 <(sort -k 1 ${GOODOTUFILE}) <(sort -k 1 ${MAPFILE}) | perl -pe "s/\\s/\\n/g" >> ${GOODSEQIDS} &&

export WORKON_HOME=$HOME/data/program_files/VirtualEnvs
source $HOME/.local/bin/virtualenvwrapper.sh
workon qiime1.9.1

filter_fasta.py -f ${INFILE} -o ${NOCHIM} -s ${GOODSEQIDS}