VIPER: Visualization Pipeline for RNA-seq, a Snakemake workflow for efficient and complete RNA-seq analysis

MacIntosh Cornwell, Mahesh Vangala, Len Taing, Zachary Herbert, Johannes Köster, Bo Li, Hanfei Sun, Taiwen Li, Jian Zhang, Xintao Qiu, Matthew Pun, Rinath Jeselsohn, Myles Brown, Xiaole Shirley Liu, Henry W. LongBMC Bioinformatics 2018 · DOI

TL;DR

This paper introduces VIPER, a user-friendly, efficient, and comprehensive RNA-seq analysis pipeline built with Snakemake. VIPER streamlines the entire RNA-seq workflow from raw data to differential expression and pathway analysis, incorporating advanced modules for immunology and other specialized analyses, making complex genomics accessible to a wider range of researchers.

What Problem Does This Paper Address?

RNA sequencing (RNA-seq) has become a cornerstone of biological research, providing deep insights into gene expression. However, the rapid evolution of RNA-seq analysis tools, coupled with the complexity of integrating them into a reproducible and scalable workflow, presents a significant challenge, especially for researchers without extensive bioinformatics expertise. Many existing pipelines are difficult to install, customize, or expand, hindering the adoption of new analytical methods. This paper addresses the need for a robust, user-friendly, and modular RNA-seq pipeline that simplifies the analysis process, integrates cutting-edge tools, and facilitates easy customization and expansion.

Methods

VIPER leverages the Snakemake workflow management system, which allows for the definition of analysis steps as interconnected "rules." This modular approach enables transparent swapping of algorithms and efficient parallelization of tasks. Key components of the pipeline include:

  • Alignment: Utilizes STAR for fast and efficient read alignment to a reference genome.
  • Quantification: Employs Cufflinks for transcript assembly and gene/isoform quantification, generating normalized read counts.
  • Quality Control: Integrates RSeQC for comprehensive assessment of read quality, including read distribution, gene body coverage, and rRNA contamination. Downsampling with Picard is used to expedite this process.
  • Differential Expression Analysis: Implements DESeq2 and Limma for robust differential gene expression testing, with DESeq2 results prioritized for downstream analysis.
  • Pathway Analysis: Incorporates Gene Ontology (GO) term analysis (using GOstats), KEGG pathway analysis (using GAGE and Pathview), and Gene Set Enrichment Analysis (GSEA) (using ClusterProfiler) to interpret the biological significance of differentially expressed genes.
  • Specialized Modules: VIPER includes optional modules for:
  • Variant Analysis: SNV calling using Varscan.
  • Fusion Gene Detection: Using STAR-Fusion.
  • Viral Analysis: Identifying viral transcripts.
  • Batch Effect Correction: Utilizing the ComBat R library.
  • Immunology Analysis: Integrating TIMER for immune cell abundance estimation and TRUST for T-cell receptor CDR3 sequence analysis.

The pipeline is configured via a human-readable YAML file and a CSV metasheet, simplifying user input.

Datasets

The paper uses a publically available paired-end RNA-seq dataset of eight B-cell acute lymphoblastic leukemia (B-ALL), three T-cell ALL (T-ALL), and three blastic plasmacytoid dendritic cell neoplasm (BPDCN) samples. This dataset is described as representing official World Health Organization (WHO) categories and is available in the NCBI Sequence Read Archive (SRA: SRP103099, BioProject ID PRJNA314605).

Key Insights and Ideas

  • Modularity is Key: The use of Snakemake is central to VIPER's flexibility, allowing for easy integration of new tools and algorithms without rewriting the entire pipeline. This is crucial in the rapidly evolving field of bioinformatics.
  • Bridging the Expertise Gap: VIPER aims to democratize RNA-seq analysis by providing a user-friendly interface and clear documentation, enabling researchers with limited computational experience to perform complex analyses.
  • Comprehensive Output: The pipeline generates informative visualizations and a consolidated HTML report, facilitating rapid understanding and communication of results.
  • Advanced Immunology Modules: The inclusion of TIMER and TRUST modules highlights VIPER's commitment to incorporating cutting-edge analyses, particularly relevant for cancer research and immunotherapy.
  • Reproducibility and Efficiency: Snakemake's ability to manage dependencies and parallelize tasks ensures efficient execution and reproducibility of analyses.

Key Results

The paper demonstrates VIPER's functionality and utility through its application to the described xenograft dataset. The results showcase:

  • Effective Clustering: Unsupervised clustering (Sample-Sample Correlation heatmap, PCA plots) successfully grouped samples based on their known biological categories (B-ALL, T-ALL, BPDCN), validating the pipeline's ability to reveal underlying data patterns.
  • Meaningful Differential Expression: Differential expression analysis identified significant gene sets related to T-cell activation and T-cell receptor signaling pathways, aligning with biological expectations for the sample types.
  • Successful Integration of Advanced Modules: The immunology modules (TIMER, TRUST) provided valuable insights into immune cell infiltration and T-cell receptor diversity within the samples.
  • Efficient Workflow: The paper highlights the speed and scalability of VIPER, noting that it can be run on multi-core servers and compute clusters.

Limitations and Future Work

  • Manual Curation for Fusion Genes: The authors note that while STAR-Fusion identifies potential fusion genes, manual curation of the top hits is recommended due to the presence of false positives.
  • Future Scalability: The authors envision implementing VIPER within an Amazon Machine Image (AMI) to enable high scalability and accessibility for users without access to local compute clusters.
  • Continuous Expansion: VIPER's modular design is intended to facilitate the ongoing incorporation of new analytical approaches and tools as they emerge.

Summary updated

BibTeX
@article{cornwell2018viper,
  author = {MacIntosh Cornwell and Mahesh Vangala and Len Taing and Zachary Herbert and Johannes Köster and Bo Li and Hanfei Sun and Taiwen Li and Jian Zhang and Xintao Qiu and Matthew Pun and Rinath Jeselsohn and Myles Brown and Xiaole Shirley Liu and Henry W. Long},
  title = {{VIPER: Visualization Pipeline for RNA-seq, a Snakemake workflow for efficient and complete RNA-seq analysis}},
  journal = {BMC Bioinformatics},
  year = {2018},
  doi = {10.1186/s12859-018-2139-9},
}
Figure