pandoc rna-cover-page.md RNASeq_part1.md -V geometry:margin=1.2in -V fontsize=12pt --listings -H rna-listings-setup.tex --latex-engine=xelatex -s -o DAY_1.pdf


#FASTQC
fastqc --help

fastqc --version

gzip -cd data/SRR1048063_1.fastq.gz | head - n 4 |less -S 

mkdir -p fastqc
fastqc --outdir=fastqc --noextract data/SRR1048063_1.fastq.gz
fastqc --outdir=fastqc --noextract data/SRR1048063_2.fastq.gz

#CREATE INDEX
cd genome
wget ftp://ftp.ensembl.org/pub/release-90/fasta/danio_rerio/dna/Danio_rerio.GRCz10.dna.chromosome.4.fa.gz
gunzip Danio_rerio.GRCz10.dna.chromosome.4.fa.gz
cd ..
less -S genome/Danio_rerio.GRCz10.dna.chromosome.4.fa 
mkdir -p genome/GRCz10_chr4

hisat2-build genome/Danio_rerio.GRCz10.dna.chromosome.4.fa genome/GRCz10_chr4/GRCz10chr4

ls -ltrh genome/GRCz10_chr4

hisat2_extract_splice_sites.py annotation/Danio_rerio.GRCz10.90.chr.gtf > annotation/ens90z10_splicesites.txt

#RUN HISAT
mkdir hisat2
hisat2 --help

hisat2 --phred33 --known-splicesite-infile annotation/ens90z10_splicesites.txt --downstream-transcriptome-assembly -p 8 -x genome/GRCz10/Drerioz10 --summary-file hisat2/SRR1048063_Aln.stat -1 data/SRR1048063_1.fastq.gz -2 data/SRR1048063_2.fastq.gz -S hisat2/SRR1048063.sam 

#SORT and BAM CONVERSION
samtools view -u hisat2/SRR1048063.sam | samtools sort -@ 8 - -o hisat2/SRR1048063.sorted.bam

samtools index hisat2/SRR1048063.sorted.bam


#RUN STRINGTIE FOR THE ASSEMBLY
stringtie --help 

stringtie hisat2/SRR1048063.sorted.bam -p 8 -o stringtie/SRR1048063.gtf -G annotation/Danio_rerio.GRCz10.90.chr.gtf 

#RUN STRINGTIE TO MERGE
stringtie -p 6 --merge -o stringtie/merged.gtf -G annotation/Danio_rerio.GRCz10.90.chr.gtf stringtie/assemblylist.txt  

#RUN GFFCOMPARE
gffcompare --help

mkdir -p gffcompare
gffcompare -r annotation/Danio_rerio.GRCz10.90.chr.gtf -s genome/Danio_rerio.GRCz10.dna.toplevel.fa -o gffcompare/gffcompare_ensembl stringtie/merged.gtf 

#RUN STRINGTIE TO RIQUANTIFY
mkdir -p stringtie_quant
stringtie -p 8 -G stringtie/merged.gtf -B -e -o stringtie_quant/SRR1048063.gtf hisat2/SRR1048063.sorted.bam

touch sample_list.txt
vi sample_list, SRR1048063 stringtie_quant/SRR1048063.gtf

mv sample_list.txt stringtie_quant

python prepDE.py -i stringtie_quant/sample_list.txt