Viral VDAP

Viral Alignment, Variant Discovery and Annotation Pipeline

Introduction

Whole-genome sequencing of pathogenic viruses has the potential to improve surveillance, classification of disease subtypes, and association of viral subtypes with disease mechanisms. In order for viral genomic data to be universally interpretable and comparable, however, best practices must be established for quality control and variant calling. This is especially challenging for gammaherpesvirus genomes, which are relatively large (>170 kb) and contain integrated human genes. Whole-genome analysis of these viruses is typically done on an ad-hoc basis by researchers working in isolation, making it difficult to know what types of comparative analyses are possible.

We have constructed a pipeline using freely available tools for quality control, alignment and SNP calling of double-stranded DNA virus paired-end short reads with the aim of providing researchers with interoperable consensus sequences and variant lists to be used for downstream analyses.

Is this the right pipeline for you?

Your virus

• Viruses that exhibit low diversity, such as gammaherpesviruses. This tool is not suitable for viruses believed to exist as quasispecies (e.g. most RNA viruses) or for calling minority variants.
• Tested on data from enriched samples which have a higher proportion of viral reads compared to metagenomics or non-cultured samples.
• Input data should consist of paired-end reads from targeted sequencing. A step to search for and remove host reads is not included in the pipeline since there should be very few host reads after targeted sequencing. Additionally, any host reads are expected to be excluded when the consensus viral sequence is built.

What you’ll need:

• FASTQ files of your NGS sequencing results.
• FASTA file of your reference genome.
• FASTA file of your adapter sequences.

If you don't have your own reference genome, follow these steps:

1. Go to https://www.ncbi.nlm.nih.gov/labs/virus/vssi/#/find-data/virus and select "Search by Virus".
2. Begin typing the name of your virus. You can use taxonomic groups (e.g., Human gammaherpesvirus...) or common names (e.g., Kaposi's sarcoma-associated herpesvirus... don’t worry, it’s an autofill, you don’t have to type the whole thing).
3. On the filter panel on the left, click "Nucleotide Sequence Type", then check "RefSeq".
4. Select the sequence you want, then download the FASTA file.

Overview of pipeline steps

• QC: Quality filtering, trimming, and minimum length filtering (Trimmomatic)

• Assembly: De novo sequence assembly (SPAdes) -> Align scaffolds to reference and condense aligned scaffolds into a consensus/draft genome (Medusa)

• Variant discovery: Compare assembly to reference genome to extract SNPs (Parsnp) [testing in progress]

• Note: On test data, de novo assembly produced a more complete assembly that better reproduced the corresponding published genome sequence than reference-based alignment. For the curious, reference-based alignment can be carried out as follows (more details below): Alignment to reference sequence (Bowtie2)-> Sequence deduplication (Samtools) -> calling variants and making consensus sequence (Samtools))

Pipeline

Trimming and filtering reads

Using Trimmomatic, we kept reads that have a minimum average quality score of 30 and a minimum length of at least 50. Reads were trimmed at either end if the bases were below a threshold quality (<3) or contained any adapter sequences. All other parameters were kept as default.

The input for this step are paired-end FASTQ files (example: jsc_1_r1.fq.gz and jsc_1_r2.fq.gz) and a FASTA file of adapter sequences (example: adapters.fa).

java -jar /path/of/Trimmomatic-0.39/trimmomatic-0.39.jar PE -threads 8 -phred33     
jsc_1_r1.fq.gz jsc_1_r2.fq.gz  
trimmed/jsc_1_forward_paired.fq.gz trimmed/jsc_1_forward_unpaired.fq.gz     
trimmed/jsc_1_reverse_paired.fq.gz trimmed/jsc_1_reverse_unpaired.fq.gz    
ILLUMINACLIP:ref/adapters.fa:2:30:10 LEADING:3 TRAILING:3 AVGQUAL:30 MINLEN:50

Parameters:

• ILLUMINACLIP: cut adapter and other illumina-specific sequences from the read.
• LEADING: cut bases off the start of a read, if below a threshold quality.
• TRAILING: cut bases off the end of a read, if below a threshold quality.
• MINLEN: drop the read if it is below a specified length.
• AVGQUAL: drop the read if the average quality is below the specified level.
• PE: paired end mode.
• -phred33: specifies the base quality encoding.
• -threads: indicates the number of threads to use.

After this step, we get 4 output files. However, we are only interested in the 2 "paired" output files where both forward and reverse reads passed the processing.

De novo assembly with SPAdes

We chose SPAdes as our de novo assembler because it can work with Illumina and Ion Torrent reads. In addition, SPAdes is capable of providing hybrid assemblies using PacBio, Oxford, Nanopore, and Sanger reads. Currently, our pipeline only takes in Illumina and Ion Torrent reads.

The SPAdes pipeline itself contains several modules:

• BayesHammer - read error correction tool for Illumina reads.
• IonHammer - read error correction tool for IonTorrent reads.
• SPAdes - iterative short-read genome assembly.
• MismatchCorrector - tool improves mismatch and short indel rates in resulting contigs and scaffold; uses BWA.

When running the following command, SPAdes will perform read correction, genome assembly, and MismatchCorrector on the 2 "paired" FASTQ files from the previous step.

spades.py --careful -1 trimmed/jsc_1_forward_paired.fq.gz -2 trimmed/jsc_1_reverse_paired.fq.gz -o assemblies/jsc_1

If you have paired-end IonTorrent data, --iontorrent will be added to the command like so:

spades.py --careful --iontorrent -1 trimmed/jsc_1_forward_paired.fq.gz -2 trimmed/jsc_1_reverse_paired.fq.gz -o assemblies/jsc_1

If you want to specify your own k-mer sizes or number of threads in the config.yaml file, they will be added to the command and will look like the following:

spades.py -k 21,33,55 -t 20 --careful -1 trimmed/jsc_1_forward_paired.fq.gz -2 trimmed/jsc_1_reverse_paired.fq.gz -o assemblies/jsc_1

Parameters:

• -o: specifies the output directory and is required.
• --careful: runs MismatchCorrector; option is recommended only for assembly of small genomes (i.e. KSHV).
• -1: file with forward reads.
• -2: file with reverse reads.
• -t (or --threads): indicates the number of threads to use; default is 16.
• -k: comma-separated list of k-mer sizes to be used (all values must be odd, less than 128 and listed in ascending order).
• --iontorrent: flag for assembling IonTorrent data.

Refine Assembly with Medusa

We selected Medusa to refine our draft genome assembly with our reference genome, GK18. Medusa can use multiple reference genomes to determine the correct order and orientation of the contigs in a graph-based approach.

java -jar ./medusa.ar -f ref/gk18.fa -i assemblies/jsc_1/scaffolds.fasta -o final_scaffolds/jsc_1_scaffold.fasta -v

Parameters:

• -i: indicates the name of the target genome file and is required.
• -o: indicates the name of the output FASTA file.
• -v: print on console how MUMmer, a package used by Medusa, is running.
• -f: indicates the path to the comparison drafts folder ~ reference genome in FASTA format

Software

Trimmomatic v.0.39

• Bolger, A. M., Lohse, M., & Usadel, B. (2014). Trimmomatic: A flexible trimmer for Illumina Sequence Data. Bioinformatics, btu170

SPAdes v3.13.0

• Bankevich A., Nurk S., Antipov D., Gurevich A., Dvorkin M., Kulikov A. S., Lesin V., Nikolenko S., Pham S., Prjibelski A., Pyshkin A., Sirotkin A., Vyahhi N., Tesler G., Alekseyev M. A., Pevzner P. A. SPAdes: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing. Journal of Computational Biology, 2012

Parameters: -k 21,33,55,77 -t 10 --only-assembler --careful

• The line listed in sbatch.sh specifies that SPAdes should run in assembly module only (--only-assembler) and applies --careful to try to reduce the number of mismatches and short indels. The k parameter refers to the k-mer sizes. We used a range of sizes from 21 to 77. The t parameter refers to the number of threads to run the software.

Medusa v1.6
scaffold.sh (All default settings) Downstream steps use the largest scaffold produced from alignment.

Parsnp v1.2

• Treangen TJ*, Ondov BD*, Koren S, Phillippy AM: Rapid Core-Genome Alignment and Visualization for Thousands of Microbial Genomes. bioRxiv (2014). doi: http://dx.doi.org/10.1101/007351

Parameters: All default settings

Snakemake v4.3.1

• Köster, Johannes and Rahmann, Sven. “Snakemake - A scalable bioinformatics workflow engine”. Bioinformatics 2012.

Parameters: All default settings

To check for errors in Snakefile or config.yaml:

snakemake -np --configfile config.yaml

To run snakemake:

snakemake --configfile config.yaml

For reference-based alignment (alternative):

Bowtie2 v2.3.5.1

• Langmead B, Salzberg S. Fast gapped-read alignment with Bowtie 2. Nature Methods. 2012, 9:357-359.

All default settings

Samtools

• 3.consensus.sh (All default settings)
• deduplication
• consensus generation: vcfutils and consensuscall-c to create a consesus sequence (FASTQ) from bowtie-produced alignments.

SeqTK

• (All default settings)
• converting FASTQ consensus sequence to FASTA format

How to use

Installation options:

Docker

The Docker image contains as well as a webserver and FTP server in case you want to deploy the FTP server. It does also contain a web server for testing the main website (but should only be used for debug purposes).

1. docker pull ncbihackathons/<this software> command to pull the image from the DockerHub
2. docker run ncbihackathons/<this software> Run the docker image from the master shell script
3. Edit the configuration files as below

DockerFile

docker pull vdap/viral_pipeline:pipeline_v1 docker run --name vdap_container -i -t -v /path/to/data vdap/viral_pipeline:pipeline_v1

comes with a Dockerfile which can be used to build the Docker image.

1. git clone https://github.com/NCBI-Hackathons/<this software>.git
2. cd server
3. docker build --rm -t <this software>/<this software> .
4. docker run -t -i <this software>/<this software>

Testing

Test Data

We tested sequences from three different KSHV cell lines with our pipeline. These sequences were derived from the Wellcome Sanger Institute and are available in the European Nucleotide Archive.

They are:

• JSC-1 - Accession number: SAMEA1709534
• BC-1 - Accession number: SAMEA1709542
• BCBL-1 - Accession number: SAMEA1709549

Here are the commands we used to download the FASTQ files for one of the samples, JSC-1, from European Nucleotide Archive in a Linux terminal:

wget ftp.sra.ebi.ac.uk/vol1/fastq/ERR244/ERR244004/ERR244004_1.fastq.gz
wget ftp.sra.ebi.ac.uk/vol1/fastq/ERR244/ERR244004/ERR244004_2.fastq.gz
wget ftp.sra.ebi.ac.uk/vol1/fastq/ERR244/ERR244022/ERR244022_1.fastq.gz
wget ftp.sra.ebi.ac.uk/vol1/fastq/ERR244/ERR244022/ERR244022_2.fastq.gz

Since there are two runs for JSC-1, we merged the Read 1 files and Read 2 files.

cat ERR244004_1.fastq.gz ERR244022_1.fastq.gz > jsc_1_r1.fq.gz
cat ERR244004_2.fastq.gz ERR244022_2.fastq.gz > jsc_1_r2.fq.gz

Reference genome

The reference genome used was the KSHV GK18 strain complete genome sequence (Accession number: AF148805). The genome for GK18 is currently the most extensively annotated KSHV sequence available, including gene and coding sequence, repeat regions, mRNA and PolyA features (Rezaee et al., 2006).

We provided a copy of the KSHV GK18 FASTA file in the ref folder.

Results

Progressive Mauve alignment of JSC-1 de novo sequence.

The newly derived JSC-1 sequence was aligned to GK18 (AF148805) and the JSC-1 genome deposited in RefSeq (GQ994937). GK18 annotations are represent as blocks, with gene regions in white and known repeat regions in red. As expected, the de novo assembly does not resolve the repeat regions.

Future Considerations

use another assembler and integrate different assemblies?

Hackathon Team

First Women-led Biodata Science Hackathon
NIH campus, Bethesda, Maryland May 8-10, 2019

Elena Maria Cornejo Castro - team lead
Eneida Hatcher - writer
Sara Jones
Sasha Mushegian - writer
Yunfan Fan - sysadmin
Rashmi Naidu

Allissa Dillman - hackathon organizer