Sequencing snakes: the BABS Genomes at SBRS 2018

This month saw the first presentation of the BABS genomes outside UNSW. A poster on the genome assemblies was presented at the Sydney Bioinformatics Research Symposium 2018 (abstract below):

Pseudodiploid pseudo-long-read whole genome sequencing and assembly of Pseudonaja textilis (eastern brown snake) and Notechis scutatus (mainland tiger snake)

Richard J Edwards, Timothy G Amos, Joshua Tang, Beni Cawood, Sabrina Rispin, Daniel Enosi Tuipulotu & Paul Waters.

Click on thumbnail for full resolution PDF. If you use anything from this work, please cite:

Edwards RJ et al. Pseudodiploid pseudo-long-read whole genome sequencing and assembly of Pseudonaja textilis (eastern brown snake) and Notechis scutatus (mainland tiger snake) [version 1; not peer reviewed]. F1000Research 2018, 7:753 (poster) (doi: 10.7490/f1000research.1115550.1)

Abstract

The precipitous drop in sequencing costs over recent years has seen the bottleneck in vertebrate whole genome sequencing (WGS) shift from data generation (sequencing) to data processing (assembly and annotation). Draft genomes generated from cheap shotgun Illumina sequencing tend to be highly fragmented with many tens of thousands of short contigs or scaffolds. This can be improved by preparing multiple paired end and “mate pair” libraries with different insert sizes, but this increases the cost of both sequencing and data storage/analysis. PacBio or Oxford Nanopore long read sequencing enables massive improvements in assembly quality but tends to be prohibitively expensive for organisms with large genome sizes, such as vertebrates. 10x Genomics Chromium “linked read” sequencing offers a solution to this problem. High molecular weight molecules of DNA are barcoded prior to standard shotgun Illumina sequencing. These barcodes can then be used for pseudo-long-read assembly, with improved handling of repetitive regions. Where heterozygous variants are dense enough, haplotypes can be phased to generate a “pseudodiploid” assembly with some regions represented as two alleles. This is all for the cost of an additional library prep with no extra sequencing. But does it work?

We have sequenced two of the deadliest venomous snakes in Australia using 10x Chromium linked reads: the mainland tiger snake (Notechis scutatus) and the eastern brown snake (Pseudonaja textilis). Supernova v2 assemblies of the data generated exceptionally high quality genomes for the price, with maximum scaffolds over 50 Mb and N50 values of 5.99 Mb for the tiger snake and 14.7 Mb for the brown snake. This was reflected in BUSCO (v2.0.1 short) completeness estimates of 87.3% (tiger snake) and 90.5% (brown snake). These data will be compared to tiger snake WGS using standard paired end Illumina NovaSeq shotgun sequencing, and discussed with respect to some of the downstream opportunities and challenges provided by pseudodiploid genome assemblies. In particular, BUSCO analysis of haploid, pseudodiploid, and non-redundant genome assemblies revealed some interesting and unexpected behaviour of this widely-used tool. We also present results from GenomeR, a Shiny app (in development) for batch kmer genome size estimation (http://shiny.slimsuite.unsw.edu.au/GenomeR/).

Snake genomes and ongoing annotation are being made available through the lab Web Apollo browser and search tool (https://slimsuite.unsw.edu.au/servers/apollo.php). We welcome contact from anyone interested in getting involved with the annotation and analysis of these genomes.

Watch this blog for more details on different aspects of the analysis.

Linked read sequencing is go!

We already have over four billion reads and 620 GB of NovaSeq Illumina data for our three tiger snakes; next week’s BABS3291 prac will look at some of the early ABySS assemblies of one of these snakes.

Phase 2 of the sequencing is now go! 10x Chromium linked read libraries were prepared at the Ramaciotti Centre for Genomics last week for one tiger snake and one eastern brown snake. These data promise to make much easier and more intact genome assemblies. We received notification today that the samples have arrived in the KCCG sequencing laboratory at the Garvan Institute for Illumina HiSeq X (“XTen”) sequencing.

Nobody knows how well linked read sequencing, which is optimised for human genomes, will work in a snake but we look forward to finding out!

Sequencing technologies used for the BABS Genome

Sequencing for the BABS Genome is being performed at the Ramaciotti Centre for Genomics at UNSW, which is one of Australia’s top sequencing centres and has a long, rich history of genome sequencing.

The Gold Standard for genome assembly is currently to combine three technologies:

1. High coverage short read sequencing for accurate base calling of unique regions.

2. Long read sequencing for assembling complex and small repetitive regions of the genome.

3. Long range sequencing for scaffolding contigs across larger repetitive regions of the genome.

We will be using a combination of three of these latest technologies for the BABS genome:

Illumina NovaSeq and HiSeq X

Short read Illumina sequencing is still the starting point for sequencing large (>0.5 Gb) genomes. Although it is impossible to assemble short read data alone into a high-quality genome, it remains the most cost-effective technology in terms of high-quality bases sequenced per dollar. Illumina sequencing struggles with regions of the genome with certain compositional bias and short read assembly fails at repetitive regions. Nevertheless, it is possible to get a useful assembly of a large portion of the “unique” genome, which includes most of the protein-coding genes.

For the 2017 BABS genome, we are using two of the latest - and most cost-effective - Illumina sequencing platform: the HiSeq X (XTen) and new HiSeq NovaSeq. These machines have a phenomenal output per run. The NovaSeq is being used for pure Illumina sequencing, whereas the HiSeq X is being used for the sequencing component of the 10X Genomics Linked Read sequencing (below).

PacBio Sequel

Whole genome sequencing and assembly has been revolutionised by the development of long read sequencing technologies by Pacific Biosciences (PacBio) and Oxford Nanopore (MinION). With typical read lengths a hundred times longer than Illumina reads, long read sequencing enables resolution of many of the shorter repetitive regions in the genome.

Long read sequencing is still comparably expensive and the budget does not stretch for a pure PacBio assembly this year. However, we will be getting some sequencing done on the new PacBio Sequel, which will help with scaffolding Illumina contigs. We also hope to be able to generate a pure PacBio mitochondrial genome; mitochondria are present in multiple copies per cell, which effectively increases the depth of coverage!

10X Genomics Chromium Linked Reads

Due to the cost (and DNA requirements) of long read sequencing, there has been considerable effort in recent years to combine cost-effective Illumina short read sequencing with additional experimental approaches to leverage long-range information. The long range service offered by the Ramaciotti Centre is 10X Genomics Chromium linked read sequencing. Unlike PacBio or MinION, this does not contiguously sequence a long DNA molecule. Instead, it uses a clever barcoding system to link short reads back to their DNA molecule of origin. 10X Genomics software then uses this linkage to regenerate pseudo-long-reads that can be used for both genome assembly and haplotype phasing.