#********************Mc_clintock********************
#!/bin/bash -x
#SBATCH --job-name=Mcclintock
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o mcclintock%a_%A.out
#SBATCH -e mcclintock%a_%A.err
#SBATCH --partition=general
#SBATCH --get-user-env=PWD

#source /opt/Modules/3.2.9/init/Modules4bash.sh
module load conda
source activate /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_mc_clintock
mkdir logs
echo 'start_retroseq' >> logs/temp_time.runtimes;
/bin/date >> logs/temp_time.runtimes;

#TE-locate TEMP  retroseq PoPoolationTE NGStemapper
run_dir=`pwd`

cd ..
bash /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_mc_clintock/mcclintock-master/mcclintock.sh -m "temp telocate retroseq popoolationte ngstemapper" -r $run_dir/sacCer2.fasta -c $run_dir/sac_cer_TE_seqs.fasta -g $run_dir/reference_TE_locations.gff -t $run_dir/sac_cer_te_families.tsv -1 /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/20x/20x_1.fastq -2 /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/20x/20x_2.fastq -p 8 -M 64;
echo 'endtime' >> logs/temp_time.runtimes;
/bin/date >> logs/temp_time.runtimes;

#********************PopoolationTE2********************
#!/bin/bash -x
#SBATCH --job-name=popoolationTE2
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o popte2%a_%A.out
#SBATCH -e popte2%a_%A.err
#SBATCH --get-user-env=PWD

module load wublast
module load hmmer
module load trf
module load repeatmasker 
module load bwa
module load samtools
bwa index LTR.temergedref.fasta;
bwa bwasw -t 3 LTR.temergedref.fasta ../../../RAW_READS/MH63/illumina-reads/reads_1_5x.fq > 5x_1.sam &
bwa bwasw -t 3 LTR.temergedref.fasta ../../../RAW_READS/MH63/illumina-reads/reads_1_5x.fq > 5x_2.sam ;
java -jar ../popte2.jar se2pe --fastq1 ../../../RAW_READS/MH63/illumina-reads/reads_1_5x.fq --fastq2 ../../../RAW_READS/MH63/illumina-reads/reads_2_5x.fq  --bam1 5x_1.sam --bam2 5x_2.sam --sort --output 5x_LTR.sort.bam;
mkdir 5x;
java -jar ../popte2.jar identifySignatures --ppileup 5x_LTR_different_coverage.ppileup.gz --mode separate --output 5x/5x_different_coverage.signatures --min-count 3;
java -jar ../popte2.jar frequency --ppileup 5x_LTR_different_coverage.ppileup.gz --signature quick/different_coverage.signatures --output 5x/5x_different_coverage.freqsig; 
java -jar ../popte2.jar pairupSignatures --signature 5x/5x_different_coverage.freqsig --ref-genome /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_popte2/LTR/LTR.temergedref.fasta --hier /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_popte2/LTR/ltr.tehier --min-distance -200 --max-distance 300 --output 5x/5x_different_coverage.teinsertions;
mkdir 5x/filter;
java -jar ../popte2.jar filterSignatures --input 5x/5x_different_coverage.freqsig --output 5x/filter/5x_different_coverage.freqsig --max-otherte-count 2 --max-structvar-count 2 

java -jar ../popte2.jar pairupSignatures --signature 5x/filter/5x_different_coverage.freqsig --ref-genome /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_popte2/LTR/LTR.temergedref.fasta --hier /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_popte2/LTR/ltr.tehier --min-distance -200 --max-distance 300 --output filter/5x_different_coverage.teinsertions

#********************Relocate2********************

#!/bin/bash -x
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH --output=relocate2.output
#SBATCH --job-name="relocate2"
#SBATCH --partition=general
#source /opt/Modules/3.2.9/init/Modules4bash.sh

#source /opt/Modules/3.2.9/init/Modules4bash.sh
module load conda
source activate /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_relocate2
#TE-locate TEMP  retroseq PoPoolationTE


python ../scripts/relocaTE2.py -c 16 -s 273 --te_fasta sac_cer_TE_seqs.fasta --genome_fasta sacCer2.fasta --fq_dir reads/40x/ --outdir LTR_output/40x/ --reference_ins sacCer2.fasta.ori.out --run

#********************Teflon********************
#!/bin/bash 
#SBATCH --job-name=teflon_10x
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o run_time_teflon.%j.%N.out
#SBATCH -e run_time_teflon.%j.%N.err
#SBATCH --get-user-env=PWD

module load samtools;
module load bwa;
module load conda;
module load trf;
module load hmmer;
module load wublast;
module load repeatmasker;
mkdir logs

source activate /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old;
python teflon_prep_annotation.py  -a prova/sample_reference_TE_locations.bed -t prova/sample_TE_hierarchy.txt  -f prova/sample_reference_TE_annotation.fasta  -g prova/sample_reference_genome.fasta -p mites; 
python teflon_prep_custom.py -e  /opt/RepeatMasker/RepeatMasker-4.0.5/RepeatMasker -g prova/sample_reference_genome.fasta -l /075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/prova/sample_reference_TE_annotation.fasta -p mites_prova
echo 'bwa_mem' >> logs/teflon_time.runtimes;
/bin/date >> logs/teflon_time.runtimes;

bwa index /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_MP/mites.mappingRef.fa;


bwa mem -t 4  -Y /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_MP/mites.mappingRef.fa /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/samples/10x/10x_1.fastq /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/samples/10x/10x_2.fastq > prova/10x_sample.sam; 
samtools view -Sb prova/10x_sample.sam | samtools sort -@ 4  -o prova/10x_sample.sorted.bam;
rm prova/10x_sample.sam;

samtools index prova/10x_sample.sorted.bam;

echo 'teflon_start' >> logs/teflon_time.runtimes;
/bin/date >> logs/teflon_time.runtimes;

python /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/teflon.v0.4.py -w /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/10xmites -d /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_TF/ -s /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/prova/sample_file_mites_10x.txt  -i 10x_sample -eb /opt/bwa/bwa -es /opt/samtools/samtools-1.3/bin/samtools -l1 family -l2 family -q 10 -sd 2488 -t 8;

python teflon_collapse.py -w /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/10xmites -d /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_TF/ -s /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/prova/sample_file_mites_10x.txt -es /opt/samtools/samtools-1.3/bin/samtools -n1 2 -n2 2 -q 10 -t 8;


python teflon_count.py -w /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/10xmites -d /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_TF/ -s /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/prova/sample_file_mites_10x.txt  -es /opt/samtools/samtools-1.3/bin/samtools -i 10x_sample -eb /opt/bwa/bwa -l2 family -q 10 -t 8;

python teflon_genotype.py -w /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/10xmites -d  /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/mites.prep_TF/ -s /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/TEFLON/TEFLoN-master/prova/sample_file_mites_10x.txt -dt pooled;

echo 'teflon_end' >> logs/teflon_time.runtimes;
/bin/date >> logs/teflon_time.runtimes;

#********************Jitterbug********************
#!/bin/bash 
#SBATCH --job-name=jitterbug
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o jitterbug.%j.%N.out
#SBATCH -e jitterbug.%j.%N.err
#SBATCH --get-user-env=PWD
module load bwa
module load samtools
module load conda
mkdir logs
echo 'bwa_aln' >> logs/jitterbug_time.runtimes
/bin/date >> logs/jitterbug_time.runtimes



bwa aln -t 12 -n 4 -o 1 -e 3 -f /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed1_10x.fq.sai /scratch/075-melo-TEmovement/RAW_READS/index_files/Nipponbare_reference.fasta  /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed1_10x.fq &

bwa aln -t 12 -n 4 -o 1 -e 3 -f /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed2_10x.fq.sai /scratch/075-melo-TEmovement/RAW_READS/index_files/Nipponbare_reference.fasta /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed2_10x.fq ;

bwa sampe /scratch/075-melo-TEmovement/RAW_READS/index_files/Nipponbare_reference.fasta /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed1_10x.fq.sai /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed2_10x.fq.sai /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed1_10x.fq /scratch/075-melo-TEmovement/RAW_READS/MH63/illumina-reads/final_renamed2_10x.fq > 10x_MH63.sam;

samtools view -bS 10x_MH63.sam > 10x_MH63.bam;
samtools sort 10x_MH63.bam > 10x_MH63_sorted.bam;
samtools index 10x_MH63_sorted.bam;
rm 10x_MH63.sam;
rm 10x_MH63.bam;

source activate /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old;
echo 'jitterbug' >> logs/jitterbug_time.runtimes
/bin/date >> logs/jitterbug_time.runtimes

python /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/jitterbug/jitterbug.py --psorted 10x_MH63_sorted.bam -t /scratch/075-melo-TEmovement/BENCHMARK/rice_annotations/Nip_MITE_filtered_orthology.gff3 -l LTRJB_10x -n 6 -o LTRJB_10x  ;

python /scratch/075-melo-TEmovement/ENVIRONMENTS/conda_jitterbug_old/jitterbug/jitterbug_filter_results_func.py -g LTRJB_10x.TE_insertions_paired_clusters.gff3 -c LTRJB_10x.filter_config.txt -o LTRJB_10x.TE_insertions_paired_clusters-QFILTERED.gff3 > logs/LTRJB_10x.jbfilterstats.txt;

echo 'end_time' >> logs/jitterbug_time.runtimes
/bin/date >> logs/jitterbug_time.runtimes

#********************ITIS********************

#!/bin/bash -x
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o itis.%j.%N.out
#SBATCH -e itis.%j.%N.err
#SBATCH --job-name="5xmitesItis"
#SBATCH --partition=general
module load conda;
source activate /scratch/075-melo-TEmovement/ENVIRONMENTS/tools_itis/;
module load R;
perl itis.pl -g MITES/IRGSP-1.0_genome.fasta -t MITES/mite_TE.fasta -l 100 -N 5xmites -1 /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/5x/final_renamed1_5x.fq -2 /scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/5x/final_renamed2_5x.fq -e Y;

#********************MELT********************
#! /bin/bash +x

#SBATCH --job-name=MELT
#SBATCH --ntasks=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o melt.PB.slurm.out
#SBATCH -e melt.PB.slurm.err
#SBATCH --get-user-env=PWD

module load bowtie/2.3.3
module load conda
source activate /scratch/075-melo-TEmovement/BENCHMARK/melt

GENOME_PATH='/scratch/075-melo-TEmovement/BENCHMARK/genomes/Nip.fna'
FASTA_INDEX='/scratch/075-melo-TEmovement/BENCHMARK/genomes/Nip.fna.fai'
BAM_FILE=/scratch/075-melo-TEmovement/RAW_READS/alignments/MH63_vs_Nip_5X_sorted.bam

MELT_EXECUTABLE='/scratch/075-melo-TEmovement/BENCHMARK/melt_executable/MELT.jar'
#MEI_LIST=/scratch/075-melo-TEmovement/BENCHMARK/CONSENSUS/NipMEI/Nip.mei.list5
MEI_LIST=/scratch/075-melo-TEmovement/BENCHMARK/CONSENSUS/NipMEI/Nip.mei.list1
GENE_ANNOTATION=/scratch/075-melo-TEmovement/BENCHMARK/00_RUNS/TEST_RUN_FABIO/melt_firstrun/files/the_gene_thing/Nip.genes.bed


ln -s '/scratch/075-melo-TEmovement/BENCHMARK/melt_executable/MELT.jar'

java -jar MELT.jar Single \
    -a \
    -c 8 \
    -h '/scratch/075-melo-TEmovement/BENCHMARK/genomes/Nip.fna' \
    -bamfile /scratch/075-melo-TEmovement/RAW_READS/alignments/MH63_vs_Nip_5X_sorted.bam \
    -n /scratch/075-melo-TEmovement/BENCHMARK/00_RUNS/TEST_RUN_FABIO/melt_firstrun/files/the_gene_thing/Nip.genes.bed \
    -t /scratch/075-melo-TEmovement/BENCHMARK/CONSENSUS/NipMEI/Nip.mei.list1 \
    -w results_5x_new

#********************TRACKPOSON********************
#!/bin/bash -l

#SBATCH --nodes=1
#SBATCH --mem=64G
#SBATCH --cpus=8
#SBATCH -o trackposon.out
#SBATCH -e trackposon.err
#SBATCH --job-name="Trackposon"

#module load conda;
#source activate bioperl;
module load bowtie
module load samtools
module load perl
module load blast
file='MH63_40X'
DB='Oryza1fam66_259'
DIR='./'

fq1=/scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/5x/final_renamed1_5x.fq
fq2=/scratch/075-melo-TEmovement/ENVIRONMENTS/t-lex/5x/final_renamed2_5x.fq
out=$file
te=$(echo $DB | awk -F "_" '{print $1}')

cd $DIR

#Build a index with the TE reference fasta
bowtie2-build Oryza1fam66_259.fa Oryza1fam66_259

#Mapping reads against TE index
bowtie2 --time --end-to-end  -k 1 --very-fast -p 6 -x $DB \
  -1 $fq1 -2 $fq2 \
  | samtools view -bS -@ 2 - > "$out"-vs-"$te".bam

#Keep unmap reads - flag unmap/map
samtools view "$out"-vs-"$te".bam | awk -F "\t" '{if ( ($1!~/^@/) && (($2==69) || ($2==133) || ($2==165) || ($2==181) || ($2==101) || ($2==117)) ) {print ">"$1"\n"$10}}' > $out-vs-$te.fa

#blast fa against IRGSP1.0 for identification of TE insertion point
blastn -db IRGSP-1.0_genome.fasta -query $out-vs-$te.fa -out $out-vs-$te.fa.bl -num_threads 8 -evalue 1e-20

#parse blast to find insertion point in unique region
perl find_insertion_point.pl $out-vs-$te.fa.bl $out-vs-$te

#sort bed 
sort -k1,1 -k2,2n $out-vs-$te.bed > $out-vs-$te.sort.bed

#coveragebed
bedtools coverage -counts -nonamecheck \
  -a IRGSP_500bp.bed -b $out-vs-$te.sort.bed \
  | awk -F "\t" '{if ($4>=2){print $0}}' > coveragebed_$out-vs-$te\_per500pb.bed
#  -a IRGSP-1.0_10kbpwindows.bed -b $out-vs-$te.sort.bed \
 # | awk -F "\t" '{if ($4>=2){print $0}}' > coveragebed_$out-vs-$te\_per10kb.bed


#cleaning temporary files
#rm $out-vs-$te.bam
#rm $out-vs-$te.fa*
#rm $out-vs-$te.bed