## Raw files at: /home/janmee/Aiptasia/0_raw_files

########################################################################################
#### TRIMMOMATIC # version 0.36 ########################################################
########################################################################################

	for i in AI_1_5197_A1M_ATCACG_L008_R AI_1_5198_A2M_CGATGT_L008_R AI_1_5199_A3M_TTAGGC_L008_R AI_1_5200_A4M_TGACCA_L008_R AI_1_5201_I1M_ACAGTG_L008_R AI_1_5202_I2M_GCCAAT_L008_R AI_1_5203_I3M_CAGATC_L008_R AI_1_5204_I4M_ACTTGA_L008_R AI_1_5205_S1M_GATCAG_L008_R AI_1_5206_S2M_TAGCTT_L008_R AI_1_5207_S3M_GGCTAC_L008_R AI_1_5208_S4M_CTTGTA_L008_R;do java -jar /home/janmee/toolbox/Trimmomatic-0-2.36/trimmomatic-0.36.jar PE -threads 10 -phred33 -trimlog /dev/null /home/janmee/Aiptasia/0_raw_files/${i}1_001.fastq.gz /home/janmee/Aiptasia/0_raw_files/${i}2_001.fastq.gz /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_paired.R1.fq /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_unpaired.R1.fq /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_paired.R2.fq /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_unpaired.R2.fq ILLUMINACLIP:all_adapter_combined.fa:4:25:9 HEADCROP:6 LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 MINLEN:35;done

# numbers of sequences
for i in *fq;do echo ${i} && grep -c ^"@HWI" ${i};done >> number_seqs.txt
## See table 1 for results

########################################################################################
#### fastq-mcf # version 1.04.803 ######################################################
########################################################################################

mkdir adapter_removal

for i in AI_1_5197_A1M_ATCACG_L008_R AI_1_5198_A2M_CGATGT_L008_R AI_1_5199_A3M_TTAGGC_L008_R AI_1_5200_A4M_TGACCA_L008_R AI_1_5201_I1M_ACAGTG_L008_R AI_1_5202_I2M_GCCAAT_L008_R AI_1_5203_I3M_CAGATC_L008_R AI_1_5204_I4M_ACTTGA_L008_R AI_1_5205_S1M_GATCAG_L008_R AI_1_5206_S2M_TAGCTT_L008_R AI_1_5207_S3M_GGCTAC_L008_R AI_1_5208_S4M_CTTGTA_L008_R;do echo ${i};
fastq-mcf -l 35 --qual-mean 25 -R /home/janmee/toolbox/Trimmomatic-0-2.36/adapters/all_adapter_combined.fa -o /home/janmee/Aiptasia/1_trimmomatic/adapter_removal/${i}_trimm_PE_adap_rem_R1.fq -o /home/janmee/Aiptasia/1_trimmomatic/adapter_removal/${i}_trimm_PE_adap_rem_R2.fq /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_paired.R1.fq /home/janmee/Aiptasia/1_trimmomatic/${i}_trimm_paired.R2.fq;
done

########################################################################################
#### BB_Split script # from BBmap # version 36.47 ######################################
########################################################################################

for i in AI_1_5197_A1M_ATCACG_L008_R AI_1_5198_A2M_CGATGT_L008_R AI_1_5199_A3M_TTAGGC_L008_R AI_1_5200_A4M_TGACCA_L008_R AI_1_5201_I1M_ACAGTG_L008_R AI_1_5202_I2M_GCCAAT_L008_R AI_1_5203_I3M_CAGATC_L008_R AI_1_5204_I4M_ACTTGA_L008_R AI_1_5205_S1M_GATCAG_L008_R AI_1_5206_S2M_TAGCTT_L008_R AI_1_5207_S3M_GGCTAC_L008_R AI_1_5208_S4M_CTTGTA_L008_R;
do /home/carol/RADseq/tools/bbmap/bbsplit.sh ref=/home/janmee/reference/aiptasia_genome.scaffolds.fa,/home/janmee/reference/phix174_ill.ref.fa,/home/janmee/reference/phix_adapters.fa,/home/janmee/reference/symbB.v1.0.genome.fa,/home/janmee/reference/28S_NCBI_sea_anemones.fa in1=/home/janmee/Aiptasia/1_trimmomatic/adapter_removal/${i}_trimm_PE_adap_rem_R1.fq.gz in2=/home/janmee/Aiptasia/1_trimmomatic/adapter_removal/${i}_trimm_PE_adap_rem_R2.fq.gz qin=33 qout=33 rebuild=t outu1=/home/janmee/Aiptasia/2_1_no_RNA_from_QC/${i}_cleaned_PE_R1.fq outu2=/home/janmee/Aiptasia/2_1_no_RNA_from_QC/${i}_cleaned_PE_R2.fq refstats=/home/janmee/Aiptasia/2_1_no_RNA_from_QC/stats/${i}_refstats.txt scafstats=/home/janmee/Aiptasia/2_1_no_RNA_from_QC/stats/${i}_scafstats.txt threads=15 local=T minid=0.7;
done


########################################################################################
#### CLARK # version 1.2.3 #############################################################
########################################################################################

./set_targets.sh /home/janmee/reference/CLARK_bacteria_virus/ bacteria viruses

for i in AI_1_5197_A1M_ATCACG_L008_R AI_1_5198_A2M_CGATGT_L008_R AI_1_5199_A3M_TTAGGC_L008_R AI_1_5200_A4M_TGACCA_L008_R AI_1_5201_I1M_ACAGTG_L008_R AI_1_5202_I2M_GCCAAT_L008_R AI_1_5203_I3M_CAGATC_L008_R AI_1_5204_I4M_ACTTGA_L008_R AI_1_5205_S1M_GATCAG_L008_R AI_1_5206_S2M_TAGCTT_L008_R AI_1_5207_S3M_GGCTAC_L008_R AI_1_5208_S4M_CTTGTA_L008_R;
do ./classify_metagenome.sh -P /home/janmee/Aiptasia/2_1_no_RNA_from_QC/${i}_cleaned_PE_R1.fq /home/janmee/Aiptasia/2_1_no_RNA_from_QC/${i}_cleaned_PE_R2.fq -R /home/janmee/Aiptasia/3_bacVir_no28/${i} -m 0 -n 15;
./estimate_abundance.sh -D /home/janmee/reference/CLARK -F /home/janmee/Aiptasia/3_bacVir_no28/${i}.csv > /home/janmee/Aiptasia/3e_bacVir_no28/abundance/${i}_abundance_vir.csv;
done