jakeyeung / mirna-pipeline

Computational pipeline for analyzing miRNA

miRNA pipeline

June 6 2014

This is a repository containing scripts used to process miRNA data.

Directory mirna_analysis contains python and R scripts used for processing miRNA. These scripts should be somewhat general and can be used for other miRNA datasets or even gene expression data sets.

Main directory contains qsub scripts I wrote to run scripts in mirna_analysis (or just programs such as bwa.

submission_scripts directory contains executable scripts that qsub can call.

Example Workflows

• Uncompressing and aligning miRNA data
• BWA workflow
• DESeq workflow
• Get target score information for a given miRNA
• Plot expression differences between an miRNA and protein-coding gene

## Processing of miRNA data

Basic ideas for what uncompress_and_align_mirna.sh does:

1. Uncompress fq files
2. Uncompress clean.txt.gz files
3. Create fa files
4. Run BWA

## Running BWA BWA can also be run separately (useful if you want to run on a local server rather than a cluster).

For example bash script: see run_bwa.sh

Calculate TPM (transcripts per million mapped reads)

Procedure for calculate_tpm_mirna.py:

1. Calculate TPM for aligned miRNA reads

For example bash script: see run_calculate_tpm_mirna.sh that calls the python script.

Annotate SAM files (output from BWA) and annotate gene information.

python mirna_analysis/annotate_aligned_reads.py --help for more information regarding the script that annotates SAM files with gene information.

Example bash script: run_annotate_aligned_reads.sh

## DESeq workflow

Procedure:

1. Merge count data for samples using mirna_analysis/merge_counts_data_for_deseq.py
2. Run mirna_analysis/deseq_normalization.R from merged count data

<a name "targetscore"/>

Get target score information for a given miRNA.

Required packages: TargetScoreData

Procedure

Rscript get_target_scores.R mirna_name outfile

Example:

`Rscript get_target_scores.R hsa-miR-517a hsa-miR-517a.targetscores

Notes:

• Only a subset of miRNAs have available target scores (likely because they have yet to do the siRNA knockdown experiment).
• To get the list of available miRNAs:

library(TargetScoreData)
targetScoreMatrix <- get_precomputed_targetScores()
names(targetScoreMatrix)

## Plot expression differences between an miRNA and protein-coding gene

Procedure for mirna_analysis/plot_exprs_diff.py:

To plot expression differences for a single gene:

• python mirna_analysis/plot_exprs_diff.py -i /exprs/of/gene/across/samples.txt -g gene_of_interest -c 'Gene ID column name' -s /pairs/of/samples.txt

To plot expression differences for a two genes (to compare expression of miRNA and a gene of interest, for example):

• Second gene may be from a different file (e.g., you may process miRNA and protein-coding genes separately and therefore have two separate files for them)
• If so, specify the additional textfile as so: python mirna_analysis/plot_exprs_diff.py -i exprs_file1 -I exprs_file2 -g gene1 -G gene2 -c 'gene_colname1' -C 'gene_colname2' -s /pairs/of/samples.txt

Notes:

• Expression file should contain a column containing gene IDs, with a column name that is specified using -c.
• Expression file should contain expression across samples. The script identifies pairs of samples (e.g. tumour and adjacent benign) through the sample pairs textfile provided by -s.