Contents

• Overview
• Learning goals
• Biological Problem
• Workflow Diagrams
• Data
• Troubleshooting
• Funding
• License for Data

Overview

This module teaches you how to perform a short-read RNA-seq Transcriptome Assembly using a Nextflow pipeline.

Learning goals:

1. From a biological perspective, demonstration of the process of transcriptome assembly from raw RNA-seq data.
2. From a computational perspective, demonstration of computing using workflow management and container systems.
3. Also from an infrastructure perspective, demonstration of carrying out these analyses efficiently in a cloud environment.

Biological Problem

The combination of increased availability and reduced expense in obtaining high-throughput sequencing has made transcriptome profiling analysis (primarily with RNA-seq) a standard tool for the molecular characterization of widely disparate biological systems. Researchers working in common model organisms, such as mouse or zebrafish, have relatively easy access to the necessary resources (e.g., well-assembled genomes and large collections of predicted/verified transcripts), for the analysis and interpretation of their data. In contrast, researchers working on less commonly studied organisms and systems often must develop these resources for themselves.

Transcriptome assembly is the broad term used to describe the process of estimating many (or ideally all) of an organism’s transcriptome based on the large-scale but fragmentary data provided by high-throughput sequencing. A "typical" RNA-seq dataset will consist of tens of millions of reads or read-pairs, with each contiguous read representing up to 150 nucleotides in the sequence. Complete transcripts, in contrast, typically range from hundreds to tens of thousands of nucleotides in length. In short, and leaving out the technical details, the process of assembling a transcriptome from raw reads (Figure 2) is to first make a "best guess" segregation of the reads into subsets that are most likely derived from one (or a small set of related/similar genes), and then for each subset, build a most-likely set of transcripts and genes.

Figure 2: The process from raw reads to first transcriptome assembly.

Once a new transcriptome is generated, assessed, and refined, it must be annotated with putative functional assignments to be of use in subsequent functional studies. Functional annotation is accomplished through a combination of assignment of homology-based and ab initio methods. The most well-established homology-based processes are the combination of protein-coding sequence prediction followed by protein sequence alignment to databases of known proteins, especially those from human or common model organisms. Ab initio methods use computational models of various features (e.g., known protein domains, signal peptides, or peptide modification sites) to characterize either the transcript or its predicted protein product. This training module will cover multiple approaches to the annotation of assembled transcriptomes.

Workflow Diagram

Figure 3: Nextflow workflow diagram.
Image Source: https://nf-co.re/denovotranscript/dev/

Explanation of which notebooks execute which processes:

• Notebooks labeled 0 and 1 (Submodule_0_Glossary.md and Submodule_1_background.ipynb) cover background materials and provide a centralized glossary for both the biological problem of transcriptome assembly, as well as an introduction to workflows and container-based computing.
• Notebook 2 (Submodule_02_basic_assembly.ipynb) carries out a complete run of the Nextflow assembly workflow on a modest sequence set, producing a small transcriptome.
• Notebook 3 (Submodule_03_annotation_only.ipynb) carries out an annotation-only run using a prebuilt, but more complete transcriptome.

Data

The test dataset used in the majority of this module is a downsampled version of a dataset that can be obtained in its complete form from the SRA database (Bioproject PRJNA318296, GEO Accession GSE80221). The data was originally generated by Hartig et al., 2016. We downsampled the data files in order to streamline the performance of the tutorials and stored them in a Google Cloud/S3 bucket. The sub-sampled data, in individual sample files as well as a concatenated version of these files are available in our Google Cloud/S3 bucket.

Additional datasets for demonstration of the annotation features were obtained from the NCBI Transcriptome Shotgun Assembly archive. These files can be found in our Google Cloud/S3 bucket.

• Microcaecilia dermatophaga
  • Bioproject: PRJNA387587
  • Originally generated by Torres-Sánchez M et al., 2019.
• Oncorhynchus mykiss
  • Bioproject: PRJNA389609
  • Originally generated by Wang J et al., 2016, Al-Tobasei R et al., 2016, and Salem M et al., 2015.
• Pseudacris regilla
  • Bioproject: PRJNA163143
  • Originally generated by Laura Robertson, USGS.

Funding

Funded by NIH/NIGMS P20GM103466.

License for Data

Text and materials are licensed under a Creative Commons CC-BY-NC-SA license. The license allows you to copy, remix and redistribute any of our publicly available materials, under the condition that you attribute the work (details in the license) and do not make profits from it. More information is available here.

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License