RNASeq Quantification Pipeline

This RNASeq Pipeline aligns sequencing reads (single or paired end), sorts and indexes the alignment (.bam), counts features and conducts a differential gene expression analysis.

The pipeline uses the following four Apps Library Capsules

1. STAR Alignment

2. Sambamba Sort & Index

3. FeatureCounts

4. DESeq2

Creating Prerequisite Data Assets

Code Ocean has supplied the datasets needed to run the Pipeline on the codeocean-public-data S3 bucket. Create a Data Asset from the public S3 bucket below. For more details, see Adding a New Data Asset.

1. Example Sequencing Reads

Bucket Name: codeocean-public-data

Path: example_datasets/Normox

In this example dataset containing reads from different samples, all reads for each sample will be in a separate folder. The Pipeline will pass each folder to the downstream alignment, allowing each sample to process in parallel.

2. hg38 Annotation

Bucket Name - codeocean-public-data

Path - genomes/hg38_Annotation

Here, we assume the reads are aligning to hg38, but any reference can be used.

3. hg38 Star Index

Bucket Name - codeocean-public-data

Path - example_datasets/STAR_GRCh38_GENCODE_Release_21_Index/star_index/

In order to create a reference for a different genome, visit the STAR Generate Genome Index Capsule in the Apps Library and follow the README in order to create a compatible index for STAR.

Create a Pipeline

Attach Data Assets

• Click Manage Data Assets

• Attach STAR Index, Annotation, and Example Sequencing Reads Data Assets.

Design Matrix

The design matrix specifies metadata associated with the samples, i.e. tumor vs normal, tissue type, etc. In order to create the design matrix:

1. Create a Folder named DesignMatrix

2. Create metadata.csv with the following 3 columns:

   • Run

   • Condition

   • Batch

Run should match the prefix for the .bam file output for the sample. Condition and Batch should indicate any metadata conditions to take into account to differentiate the samples in DESeq2.

Assemble Pipeline

Add the following Code Ocean Apps Capsules to the Pipeline Builder area:

• STAR Alignment

• Sambamba Sort & Index

• FeatureCounts

• DESeq2

Configure Connections

In order to configure the connection for each step:

1. Click Settings
2. Reads Dataset to STAR Alignment is set to Default.

3. HG38 Star Index to STAR Alignment is set to Collect.

4. STAR Alignment to Sambamba Sort & Index is set to Default

5. Sambamba Sort & Index to FeatureCounts is set to Collect

6. Annotation Data Asset to FeatureCounts is set to Default

7. FeatureCounts to DESeq2 is set to Default

   • Set the destination to Counts_data

7. Create and attach DesignMatrix - metadata.csv to DESeq2.

   • Select Connection to Default, set the destination to Counts_data

Completed Pipeline

Configure App Panel

On the App Builder tab, click Create App. Click Create App again if prompted. Click Finish.

Parameters

Configure the App Panel as follows.

Reference READMEs in Capsules to find out more about the parameters used.

The DESeq2 “Design formula” is based on the columns supplied in the design matrix.

Results

• nextflow

  • Consists of logs describing actions of nextflow.

• DESeq2_results.csv

  • .csv file with the results table

• MA_plot.png

  • MA plots display a log ratio (M) vs an average (A) in order to visualize the differences between two groups. In general we would expect the expression of genes to remain consistent between conditions and so the MA plot should be similar to the shape of a trumpet with most points residing on a y intercept of 0.

• PCA.png

  • Visualize how the samples group by treatment

• volcano_plot.png

  • The volcano plot enables it to simultaneously capture the effect size and significance of each tested gene.

• plots_by_gene

  • A folder containing a file for each gene that plots the normalized counts for a single gene in order to get an idea of what is occurring for that gene across the sample cohort.