Towards Accreditation in Metagenomics for Clinical Microbiology

Thesis Comittee

Programa de Doutoramento do Centro Académico de Medicina de Lisboa 

Inês Mendes

15th of September, 2020

Metagenomics

Random "shotgun" sequencing of microbial DNA, without selecting a particular gene.

Promising methodology for obtaining fast results for the identification of pathogens and their virulence and antimicrobial resistance properties without the need for culture.

 | The motivation

Who is there? - Taxonomic identification

What are they doing? - Virulome, Resistome, Functional Annotation

Who is doing what? - Functional Assignment

Metagenomics

 | The motivation

  • Development of novel methods and metrics to accurately identify and estimate relative abundance of viral and baterial pathogens.
  • Standardizing the process of metagenomic analysis.
  • Development of a computational efficient and robust framework.

Metagenomics

 | The motivation

Main Goals

  • Fundação para a Ciência e a Tecnologia (SFRH/BD/129483/2017)
  • Dr. João A. Carriço - Instituto de Medicina Molecular, MRamirez Lab
  • Prof. Dr. John W. A. Rossen - University Medical Center Groninen, Department of Medical Microbiology

Metagenomics

 | The motivation

Funding

Promoters & Host Institutions

 | Analysis

Metagenomics

Precision, Sensibility & Performance

(Culture + Maldi-TOF)

11 Metagenomic Samples - Fluid & Tissue

 | Analysis

Metagenomics

  • Highlights the potential and the limitations of shotgun metagenomics as a diagnostic tool - Lack of reproducibility
  • Results are highly dependent on the tools, and specially database, chosen for the analysis - Lack of standardization and propper benchmark.

 | Analysis

Metagenomics

Major issues

Reproducibility in Computational Biology

Clinical Shotgun Metagenomic Analysis

Programa de Doutoramento do Centro Académico de Medicina de Lisboa 

Inês Mendes

15th of September, 2020

Reproducibility

 | The motivation

  • Do I know what is happening?
  • Is it reproducible?
  • Is it shareable?

Black

Box

Transparent

Box

  • Commercial/Freeware
  • You get what it gives you
  • Ready to use
  • Stealth change
  • Standalone
  • Freeware
  • You can "tailor"
  • "Major" headache
  • Visible change
  • Dependencies

The needs:

  • Analyze a large amount of sequence data routinely
  • Some computationally intensive steps
  • Constantly updating/adding software

Reproducibility

 | The motivation

Writing of pipelines in python/perl/shell scripts circa 2000, colorized.

  • Custom ad-hoc scripts
  • Difficult to parallelise
  • Difficult to install/run
  • Hard to deploy in multiple environments 
  • What's workflow managers?!
  • What's docker?!

Workflows in the Paleolithic era:

Reproducibility

 | The motivation

The game changing combination of workflow managers and containers:

  • Portability
  • Reproducible
  • Scalability
  • Multi-scale containerization
  • Native cloud support

Reproducibility

 | The motivation

Workflows in the Modern era:

Monolithic pipelines

Need to change often

Siloed tool containers

Don't do much by themselves

Reproducibility

 | The motivation

Reproducibility

 | 

Workflow based development

Component based development

Components are modular pieces with some basic rules:

Component A

- Input/Output

- Parameters

- Resources

Component B

- Input/Output

- Parameters

- Resources

Reproducibility

 | 

With this framework, building workflows becomes simple:

flowcraft build -t 'trimmomatic fastqc spades pilon' -o my_nextflow_pipeline

Results in the following workflow DAG (direct acyclic graph)

It's easy to get experimental:

flowcraft build -t 'trimmomatic fastqc skesa pilon' -o my_nextflow_pipeline

Switch spades for skesa

Reproducibility

 | 

Forks

Connect one component to multiple

Secondary channels

Connect non-adjacent components

Extra inputs

Inject user input data anywhere

Recipes

Curated and pre-assembled pipelines for specific needs

Reproducibility

 | 

Reproducibility

 | 

Reproducibility

 | 

Multiple Raw Input Types

Not limited to paired-end FastQ or Fasta

Dynamic Input in Components

One component, multiple inputs

Expand Building Features

New merge operators

  • Component that receives multiple input types
  • Component that merges multiple inputs from different branches

Reproducibility

 | 

Programa de Doutoramento do Centro Académico de Medicina de Lisboa 

Inês Mendes

15th of September, 2020

dengue virus genotyping from amplicon and shotgun metagenomic sequencing

 | The motivation

doi:10.1038/nrmicro1690

Sequential infection increases the risk of a severe form of the infection - dengue hemorrhagic fever.

 | The motivation

Dengue hemorrhagic fever:

  • The virus presents 4 different serotypes, sub-classified in many genotypes
  • After "primary" infection with one serotype, "secondary" infections by one or more of the other serotypes can precipitate ‘antibody dependant enhancement’ (ADE)
  • Infection with one type gives little immune protection against the other types

 | The motivation

  • DENV-1  - genotypes I-V​​
  • DENV-2 - genotypes Asian I, Asian II, Cosmopolitan, American, Asian/American & Sylvatic​
  • DENV-3 - genotypes I-V​
  • DENV-4 - genotypes I - III & Sylvatic​

https://doi.org/10.1371/journal.pntd.0001876.g002

https://doi.org/10.1371/journal.pntd.0000757

Thailand

Viet Nam

 | The motivation

DENV: (+)ssRNA (~11Kb; 1 ORF)​

The single polyprotein encodes:​

  • Structural Proteins:​

    • C – capsid​

    • prM – pre-membrane​

    • M - membrane

    • E - envelope​

  • Non-Structural Proteins: ​

    • NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5​​

 | The motivation

Empower the use of HTS to monitor the dissemination of the disease

RNA Extraction

 PCR
Amplification

HTS Sequencing

➔

➔

How do the different genotypes model transmission and infection?

 | The motivation

 | The workflow

  • Ready to use but customizable
  • Scalable
  • Reproducible
  • Stand-alone (confidential data)
  • Easy to explore and share results

Requirements

A solution

 | The workflow

DENV Identification

  • 3830 complete DENV genomes from the NIAID Virus Pathogen Database and Analysis Resource (ViPR)​
  • complete genome sequence ​
  • human & mosquito host (exception of DENV-1 III, monkey)​
  • collection year (1950-2019)​

In Silico Typing:​

  • 161 representative sequences of all sero and genotypes.​

 | The workflow

  • DENV-1  
  • DENV-2 

a) Envelope Region  b) whole genome sequence 

 | DENV Classification

a) Envelope Region  b) whole genome sequence 

  • DENV-3  
  • DENV-4 

 | DENV Classification

Shotgun Metagenomics dataset:

  • 9 plasma samples​
  • 13 serum samples​
  • 1 spiked sample with the 4 serotypes​
  • Positive and Negative controls​
nextflow run DEN-IM.nf -profile slurm_shifter --fastq="fastq/*_{1,2}.*"
  • HPC with 300 Cores/Processing Power and 3 TB RAM
  • On average, each sample took 7 minutes to analyse. A total of 75 CPU hours were used to analyse the 25 samples, with a total of 17Gb in size. This analysis resulted in 69Gb of data.

 | Use case

 | Use case

Git, Nextflow (java) and a container engine (Docker, singularity, shifter...).​

apt-get install git
curl -s https://get.nextflow.io | bash 

apt-install docker-ce

Clone (or run remotely)

git clone https://github.com/B-UMMI/DEN-IM.git
https://github.com/B-UMMI/DEN-IM/wiki

 | Installation

(Meta)Genomic Assembly Benchmark

 de novo Assembly of short-read data

Programa de Doutoramento do Centro Académico de Medicina de Lisboa 

Inês Mendes

15th of September, 2020

The assembly methods provide longer sequences that are more informative than shorter sequencing data and can provide a more complete picture of the microbial community in a given sample.

Benchmark

 | (Meta)Genomic assembly

Benchmark

 | (Meta)Genomic assembly

Benchmark

 | (Meta)Genomic assembly

Benchmark

 | Assembly workflow

Reference Dataset (Complete Bacterial Genomes)

In silico mock sample (even)

In silico mock sample (log)

Zymos standard (even)

Zymos standard (log)

3.7 M read pairs

8.8 M read pairs

47.8 M read pairs

Assembly Workflow

Assembly Quality Assessment

Benchmark

 | Assembly evaluation

Reference Dataset (Triple)

Assembly file (fasta)

Filter min contig size (1000 bp)

Mapping with Minimpa2

Read Data

PAF file (tab)

Benchmark

 | Assembly evaluation

General Assembly & Global Mapping Statistics

Mapping Statistics & Metrics per Reference

  • Number of contigs
  • Number of basepairs
  • N50
  • Max contig size
  • Number of contigs (%)
  • Number of basepairs (%)
  • N50
  • % Mapped contigs 
  • % Mapped basepairs
  • % Mapped reads

Original

Filtered (1000bp)

  • Contiguity
  • Identity (average)
  • Identity (min)
  • Breadth of coverage
  • C90 & C95
  • Number of aligned contigs
  • Phred quality score per contig

Benchmark

 | Assembly evaluation

C90 & C95

Number of contigs to cover at least 90% and 95% of the reference genome, respectively. 

Contig Phread Quality Score

E = 1 - Identity
Phred(E) = \begin{cases} -log(E) * 10 & \quad \text{if } E \text{< 0}\\ 60 & \quad \text{if } E \text{= 0} \end{cases}

Contiguity

Longest percentage of the reference sequence assembled in a single contig.

Benchmark

 | Performace

Benchmark

 | Mock sample (even)

Benchmark

 | Mock sample (even)

Benchmark

 | Mock sample (even)

Benchmark

 | Mock sample (even)

Bradth of coverage and number of contigs per Reference for each Assembler

Benchmark

 | Mock sample (even)

Benchmark

 | Mock sample (even)

Contig Phred Quality Score  for GATBMiniaPipeline's Pseudomonas aerugiona assemby

Contig Size

Phred Score

Phred(E) = \begin{cases} -log(E) * 10 & \quad \text{if } E \text{< 0}\\ 60 & \quad \text{if } E \text{= 0} \end{cases}

Benchmark

 | Mock sample (even)

Contig Phred Quality Score per Reference for each Assembler

Data Structures & the SARS-CoV-2 Contextual Data Specification 

Programa de Doutoramento do Centro Académico de Medicina de Lisboa 

Inês Mendes

15th of September, 2020

PHA4GE

 | Data structures workgroup

Standardized data structures and interchangable formats - critical to the development of an open software ecosystem.

Focus on the development, adaptation and standardization of data models for microbial sequence data, contextual metadata, results and workflow metrics to improve the transparency, interoperability and reproducibility of public health sequencing workflows.

Main Goal

PHA4GE

 | SARS-CoV-2 Data Spec

SARS-CoV-2 contextual data specification that incorporates publicly available community standards, as well as additional fields and guidance appropriate for public health surveillance and analyses.

  • Collection template and controlled vocabulary pick lists
  • Reference guide
  • Collection template SOP
  • Data submission protocol (GISAID, NCBI, EMBL-EBI)
  • JSON structure of PHA4GE specification

Resources

PHA4GE

 | SARS-CoV-2 Data Spec

Future Work

Continue benchmark analysis for the mock semple (log distributed) and the real samples (Zymos community standards log and evenly distributed).

PHA4GE - Harmonization of tool outputs for the detection of antimicrobial registance genes (https://github.com/pha4ge/hAMRonization).

Web service for the interactive vizualization of Kraken's taxonomic composition reports.

Nothing else Meta - Reference indenpendent filtration of human reads from (meta)genomic datasets.

Special thanks to Diogo Silva, Bruno Gonçalves, Tiago Jesus, Pedro Vila-Cerqueria, Rafael Maria Mamede, João Carriço, John Rossen and Mário Ramirez.

Thank you for your attention

Towards Accreditation in Metagenomics for Clinical Microbiology

By Inês Mendes

Made with Slides.com

Towards Accreditation in Metagenomics for Clinical Microbiology

CAML PhD Program Thesis Comittee - 15 September 2020

  • 526

More from Inês Mendes