Speakers

The management of raw nanopore sequencing data poses a challenge that must be overcome to accelerate the development of new bioinformatics algorithms predicated on signal analysis. SquiggleKit is a toolkit for manipulating and interrogating nanopore data that simplifies file handling, data extraction, visualisation, and signal processing. Its modular tools can be used to reduce file numbers and memory footprint, identify poly-A tails, target barcodes, adapters, and find nucleotide sequence motifs in raw nanopore signal, amongst other applications. SquiggleKit serves as a bioinformatics portal into signal space, for novice and experienced users alike. It is comprehensively documented, simple to use, cross-platform compatible and freely available (https://github.com/Psy-Fer/SquiggleKit).

Assemblies of small eukaryotic genomes using long-reads are often close to complete. However, these assemblies remain difficult to validate, especially when genomes have complex features such as large inversions, translocations, ploidy variations, and where chromosome number may not be known. While many tools for assessing assemblies with short-reads exist, long-reads have far greater power for confirming the accuracy and completeness of contigs. I will present Tapestry, a tool for validating the contigs of a small assembly automatically and visualising the contigs so the structure of the assembly can be refined before polishing. I will show how Tapestry has helped us to resolve the complex genomes of several small eukaryotes. 

Percent Spliced-In (PSI) values are commonly used to report alternative pre-mRNA splicing (AS) changes. Previous PSI-detection tools were limited to specific AS events and were evaluated by in silico RNA-seq data. We developed PSI-Sigma, which uses a new PSI index, and we employed actual (non-simulated) RNA-seq data from spliced synthetic genes (RNA Sequins) to benchmark its performance (i.e., precision, recall, false positive rate, and correlation) in comparison with three leading tools (rMATS, SUPPA2, and Whippet). PSI-Sigma outperformed these tools, especially in the case of AS events with multiple alternative exons and intron-retention events. We also briefly evaluated its performance in long-read RNA-seq analysis, by sequencing a mixture of human RNAs and RNA Sequins with nanopore long-read sequencers. Based on the long-read RNA-seq data of RNA sequins, we found that nanopore long-read RNA-seq is qualitatively reliable. Also, in human U87 cells, we found that ~1 million long reads can already detect major AS changes in ~3,500 protein-coding genes with at least 10 supporting long reads. PSI-Sigma is implemented in Perl and is available at https://github.com/wososa/PSI-Sigma

Recent years has seen an uptake in the use of GPUs for Genomics, from basecalling (e.g. Guppy) to variant calling (e.g. Deep Variant). Long-read sequencing technology such as Oxford Nanopore sequencing holds the promise of simple and cost-effective de novo assembly. This is important for generating reference sequences (even for complex, polyploid organisms) and identifying structural variants such as deletions and translocations. One of the difficulties however of high-quality de novo assembly is its substantial computational cost.  Post-sequencing assembly can take longer than the sequencing experiment itself. The Nvidia genomics team is harnessing the power of GPUs to develop a pipeline for massive acceleration of de novo assembly. Our end goal is real-time long-read de novo assembly.

Thanks to Oxford Nanopore Technologies, long-read sequencing is becoming more accessible for a much broader range of applications and end-users. Bioinformatics analysis was already a bottle-neck with the previous generations of sequencing technologies, but even more so with the new generations. Nanopore-based sequencing technologies are so much more accessible and can rapidly produce so much more data that the data analysis challenges can become fundamental. Community-driven solutions to democratize data analysis is crucial in the same way Oxford Nanopore is democratizing sequencing. Galaxy has been shown to be a successful option for short-read sequencing, but we think its advantages will shine even more in the era of long-read sequencing. Firstly, the user-friendly web interface does not require advanced computational skills, making it ideally suited for this interdisciplinary area and educational purposes. Secondly, the software and workflows can be seamlessly upgraded at the server side while maintaining 100% reproducibility of the performed analysis. Thirdly, the computational infrastructure supports a diverse spectrum from personal computers to cluster grids and the cloud. Within the scope of this project, we provide Oxford Nanopore-related tools in Galaxy. We have developed a collection of the best practice workflows for genome assembly within Galaxy. Our work is available for everyone at the European Galaxy server (https://usegalaxy.eu) and supportive self-learning training material is available. I will also introduce the Street Science Community (https://streetscience.community), a voluntary-based non-profit group that aims to teach the public the fundamental concepts of molecular biology and genetics data analysis by analyzing the  “DNA of beer” using MinION and Galaxy.

Once you can generate terabytes of data from a single flowcell, simply moving and storing that data can become the bottleneck for your workflow. We present some recent and up-coming MinKNOW features designed to help with your data management challenges. Find out how to trigger analyses automatically and hear some rules of thumb to help you plan what kit you need. We will also give you some insight into how Oxford Nanopore has scaled out data management to hundreds of devices.

UNCALLED is a tool that maps raw nanopore reads to large DNA references as they are being sequenced. It is a streaming algorithm, meaning the mapping begins as soon as the first bit of signal comes from the sequencer. UNCALLED can currently map reads from all active MinION pores to a 31Mbp reference containing eight bacterial genomes after less than three seconds of sequencing and analysis per read. The main application for UNCALLED is ReadUntil sequencing, where reads can be ejected from the pore depending on whether or not they map to the reference.

Once you can generate terabytes of data from a single flowcell, simply moving and storing that data can become the bottleneck for your workflow. We present some recent and up-coming MinKNOW features designed to help with your data management challenges. Find out how to trigger analyses automatically and hear some rules of thumb to help you plan what kit you need. We will also give you some insight into how Oxford Nanopore has scaled out data management to hundreds of devices.

The Bioinformatics Resource section of the Nanopore Community is an evolving repository of data analysis tutorials which aim to deliver best-practise workflows for researchers to explore their own data. The installation of software dependencies is managed through bioconda and the data analysis is orchestrated using the Snakemake workflow management system. The R Markdown package is used to merge bioinformatics code and nanopore data into a reproducible and literate document. The tutorials are packaged with example data, and the complete tutorial code is placed on our Github pages. In this data analysis session, I will introduce our tutorials and will demonstrate how they can be used by laboratory researchers developing their bioinformatics skills. Tutorials cover topics such as the basic QC of individual flowcell runs, mapping reads to a reference genome and the quantitative analysis of transcriptome data. A more technical dissection of a tutorial should illustrate how the tutorials can be modified and customised to your needs. We would welcome your feedback and requests for future tutorial topics during the Q&A at the end of the presentation.