Collection of bash scripts used throughout the pipeline

bash-scripts/download_rnaseq.sh

@@ -0,0 +1,61 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+cat accessions.txt | while read line
+do
+    echo "Reading ${line}"
+    if [ ! -f /gpfs/home/naa17jnu/scratch/seq_runs/${line}.fastq.gz ] # check if file doesnt already exist
+    then
+        echo "File ${line} doesnt exist - downloading"
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/000/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 000"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/000/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/001/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 001"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/001/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/002/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 002"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/002/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/003/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 003"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/003/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/004/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 004"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/004/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/005/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 005"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/005/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/006/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 006"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/006/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/007/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 007"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/007/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+        if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/008/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 008"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/008/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+         if wget --spider ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/009/${line}/${line}.fastq.gz 2>/dev/null
+        then
+            echo "Found ${line}.fastq.gz on 008"
+            wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR838/009/${line}/${line}.fastq.gz -P /gpfs/home/naa17jnu/scratch/seq_runs
+        fi
+    fi
+done

bash-scripts/fastqdump.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+
+source activate icshape-pipe
+fasterq-dump SRR11164866 -O /gpfs/home/naa17jnu/scratch/fastq_output

bash-scripts/runR.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+#SBATCH --mem=80G
+#SBATCH -p compute-24-96
+
+export PATH=/gpfs/home/naa17jnu/.conda/envs/r-env/bin:$PATH
+source activate r-env
+Rscript sleuth-new.R

bash-scripts/run_fastp.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+
+cat accessions.txt | while read line
+do
+    echo "Cleaning ${line}"
+    ./fastp -i /gpfs/home/naa17jnu/scratch/seq_runs/${line}.fastq.gz -o /gpfs/home/naa17jnu/scratch/fastp/${line}_clean.fastq.gz -Q -L -j ${line}.json - h ${line}.html -R "${line}"
+    echo "Done"
+done

bash-scripts/run_fastqc.sh

@@ -0,0 +1,13 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+
+
+# MUST DO module load fastqc/0.11.9 before running
+module load fastqc/0.11.9
+cat accessions.txt | while read line
+do
+    echo "Cleaning ${line}"
+    fastqc /gpfs/home/naa17jnu/scratch/seq_runs/${line}.fastq.gz --outdir /gpfs/home/naa17jnu/scratch/fastqc_analysis
+    echo "Done"
+done

bash-scripts/run_format.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+cat input_mirs.txt | while read line
+do
+    echo "Reading ${line}"
+    python miR-Search.py format -i sleuth_output/sleuth_results_${line}.csv mir_search_output/${line}.csv -o ${line}.csv
+    echo "Done"
+done

bash-scripts/run_format_shape.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+cat input_mirs.txt | while read line
+do
+    echo "Reading ${line}"
+    python mir-search ggplot-format -s "sleuth_output_transcript/condition${line} .csv" icshape-align_out_flanking/${line}.pickle -o ${line}.csv
+    echo "Done"
+done

bash-scripts/run_icshape_align.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+cat input_mirs.txt | while read line
+do
+    echo "Reading ${line}"
+    python mir-search icshape-align -s /home/sammy/Documents/MSc_Research_Project/9.transSHAPE/final.shape -t mir_search_output/${line}.csv -u final_trans.fasta -f 10
+    echo "Done"
+done

bash-scripts/run_kallisto.sh

@@ -0,0 +1,26 @@
+#!/bin/bash
+#SBATCH [email protected]
+#SBATCH --mail-type=END
+
+#Before running generate index file - kallisto index -i filename
+
+# DO module load kallisto/0.46.1
+module load kallisto/0.46.1
+line=$1
+echo "Reading ${line}"
+if [ -f /gpfs/home/naa17jnu/scratch/fastp/${line}_clean.fastq.gz ] # check clean file actually exists
+then
+echo "File exists at /gpfs/home/naa17jnu/scratch/fastp/${line}_clean.fastq.gz" 
+    if [ ! -f ${line}_block ] # check kallisto isnt in progress
+    then
+        echo "Running kallisto on ${line}"
+        touch ${line}_block
+        kallisto quant -i /gpfs/home/naa17jnu/scratch/kallisto/homo_sapiens_GRCh38 --single --fragment-length=180 --sd=20 --bias -b 100 -o /gpfs/home/naa17jnu/scratch/kallisto/${line}_kallisto /gpfs/home/naa17jnu/scratch/fastp/${line}_clean.fastq.gz 2>&1 | tee ${line}_stdout.txt
+    else
+        echo "Block file found - is kallisto running already?"
+    fi
+else
+    echo "Cleaned file not found"
+fi
+rm -f ${line}_block
+done

bash-scripts/run_mir_search.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+cat input_mirs.txt | while read line
+do
+    echo "Reading ${line}"
+    python mir-search mir-search -m -i ${line} final_trans.fasta
+    echo "Done"
+done