Outline

• Program overview

• Program publications

• Program science

  • Viral discovery and machine learning
  • Historical reconstruction with metagenomics

• Improvements to user science

  • Automating Stable Isotope Probing (SIP) metagenomics
  • Community profiling with iTags
  • metagenome assembly and binning
  • Global metagenome comparisons

Interconnected 

Function driven 

Genome centric 

Metagenome program overview

Metagenome program overview

• Data Products

• Community  iTags         10,000

• Metagenomes                         825

• Metatranscriptomes           900

User driven science

Program driven science

• Metagenome program science
• Viral discovery and function
•  Historical reconstruction
• Machine learning for metagenomics

• Microbial systems group science
• Plant microbe interactions
• Wetland biogeochemistry

Metagenome program in context

Carbon Cycling

Biofuels

Biogeochemistry

Integrated, system based approaches to understanding:

• Development of cross-platform analyses
• Integration of data across projects
• Research into carbon cycling and plant-microbe interaction

Metagenome program projects

Metagenome program publications

Metagenome program publications

Discovery of canidate radiation phyla

• A large group of uncultivated phyla are sound in groundwater
• Small cells, minimal genomes
• The phyla have highly divergent 16S genes with introns

Metagenome program publications

Salicylic acid modulates root colonization

The discovery of a mechanism for allowing colonization by endophytic bacteria

Metagenome program publications

Methane production in restored wetlands

Does restoring wetlands help or hurt climate change?

Methanogen abundance and methane emissions from new wetlands are dependent on electron acceptors, salinity and age

Metagenome program science

Viruses as ecosystem drivers

Viral discovery in metagenomes and metatranscriptomes by machine learning

The first soil virus metagenome

The first "complete" virus metagenome (single and double stranded DNA and RNA viruses)

Diel infection of RNA viruses in lakes

Finding highly divergent RNA viruses

Higher information content 

Less bias

• Viral classifier with 95% recall
• Increasing precision by incorporating homology information to identify known organisms
• Screening of  all metatranscriptomic data from IMG and Tara Oceans

• Machine learning applications

• Supervised

• Supervised

• Unsupervised

• Unsupervised

Classification in metagenomics

GeneLearn

A modular application for machine learning from sequence data

Construction 1649

British seige 1803

Cholera 1853

Cholera sequence

Improvements to user science

Interconnected

Genome centric

Function driven

iTags overhaul

Genome binning

Improved assembly

SIP ETOP

Host DNA depletion

Expression analysis

MT insert size

Gaia assembler 

Interconnected 

Function driven 

Genome centric 

Automating stable isotope probing

Stable isotope probing (SIP) is a method to identify the genes of microbes using a specific compound

SIP has been too complicated and time consuming to be widely adopted.  This ETOP  simplifies SIP to make it more widely available to JGI users

Jennifer Pett-Ridge

unlabeled DNA

labeled DNA

SIP automation ETOP

SIP automation ETOP

Current SIP-’omics approach is low-throughput

    and requires special equipment

LLNL approach will use NanoSIMS for sample

    prescreening prior to SIP processing/sequencing

Will also improve density gradient separation with

    intercalators

And will automate:

• Fraction collection

• Density profile Characterization

• Fraction cleanup (desalting)

• Nucleic acid quantification

• Reverse transcription and amplification 

Total hands-on processing

Itags amplicon sequencing 

25,000 Itags sequenced since 2013

Itags are a useful, cost effective way of profiling communities  but they are not being fully used.

Phase 1: Sequence the V4 and V5 region

Phase 2: Integration of all itag data and enhanced analysis

• Open Reference OTU picking across all JGI Itag data
• Improved metadata search and visualization tools 
• More analyses e.g. Bayesian ecological networks

Metagenome assembly and binning

• Complete overhaul of Metagenome and Metatranscriptome assembly reduced resource use and increased assembly quality

• Publication of Metabat  for automatic genome binning
• Publication of Elviz for manual binning and exploration

All vs. all metagenomic assembly

• Metagenomic data are only compared against  reference reads
• The cost of recomputing annotations when new references are added is high

The challenge:

The Gaia Assembler is a global, distributed, asynchronous assembly and alignment program designed to continuously align and annotate read data of arbitrarily large size. 

Host Genome depletion

The challenge:

Endophytic bacteria cannot be effectively sequenced because host DNA overwhelms the sample

Working with a commercial partner to develop  host specific depletion probes from genomic DNA

Charge Questions

Should the metagenome program prioritize functional methods like SIP and metatranscriptomics over genome reconstruction improvements?

Should we begin directing more resources towards developing methods to reconstruct metabolic networks?

Should more work be done to make metagenome data available in a machine readable way, even if  that means that fewer interactive tools are available?