Using Global Metagenomes to Quantify PCR Primer Coverage and Integrating 3-Domain Amplicon Data with Models

Jesse McNichol¹, Paul M. Berube², Steven J. Biller^2,3,

Sallie W. Chisholm², Jed A. Fuhrman¹

 

¹University of Southern California

²Massachusetts Institute of Technology

³Wellesley College

 

• Universal Ribosomal PCR Primers target the SSU rRNA of (almost) everything:
  • Quantify the whole cellular community from A-Z (Archaea + Bacteria  + Organelles + Protists + Zooplankton)
  • 16S and 18S on same scale - ratios meaningful
  • Low cost (~$8/sample, 200,000 sequences/ea)
  • Amplicon sequence variant (ASV) methods allow intercomparison and mapping

Universal primers as a mapping tool

 

• Detailed protocols available on https://fuhrman-lab.github.io
  • Lab protocol
  • in-silico workflow:
    • Splits 16S and 18S automatically (analyzed separately)
    • Denoises with DADA2/deblur in qiime2 => ASVs
    • Splits data automatically into categories (e.g. chloroplasts)
    • Step-by-step guide on protocols.io

Universal primers are easy to analyze!

Universal primers complement existing techniques

Universal primers complement existing techniques

 

• Simons Collaborative Marine Atlas Project (cmap.readthedocs.io)
• Unify data retrieval and co-localization
• ASVs are stable, and make sense in database

Universal primers as a mapping tool - CMAP!

 

 

1. Metagenomic validation of universal primer quantification and environmental coverage
2. Use cases for universal primer data (with BioGEOTRACES data)
3. Future work

Talk Outline

Mock communities show PCR can be quantitative

Similarly quantitative "in the wild"

Similarly quantitative "in the wild"; 18S less so

Similarly quantitative "in the wild"; 18S less so

PCR amplicon

abundances (18S)

What about primer bias?

MGPrimerEval : a reproducible snakemake pipeline

• Compares primers to unassembled MG reads
• Reproducible by conda/docker integration
• tinyurl.com/titus-workflow-philosophy
• github: tinyurl.com/MGPrimerEval

https://www.physalia-courses.org

MGPrimerEval : datasets

• TARA oceans (458), BioGEOTRACES (610), MBARI bloom MG (84), Malaspina FL/PA samples (58), SPOT/Catalina MG (8)
  • Total = 1,218 metagenomes
  • Globally distributed, mostly oxic, pelagic samples

Pesant et al., 2015

Sieradzki et al., 2019

Acinas et al., 2019

Biller et al., 2018

MGPrimerEval : predicted coverage (0-mismatch)

MGPrimerEval : conclusions

• 515Y/926R universal primers have best coverage overall
  • Nearly 100% for 16S, ~90% for 18S
    • 18S has similar coverage vs. Euk-specific primers
  • Will provide accurate community profiles across pelagic ocean regions
  • Some rare taxa have low coverage
• Some other primers have unexpectedly low coverage
• Improvements can be made to any primer to:
  • Maximize overall coverage across diverse environments
  • Optimize to a particular environment

Use cases: BioGEOTRACES amplicons

NB: 18S usually small fraction of total

3. Use cases:

 

Resolution

• Are short amplicons (~373 bp) enough to resolve ecologically-relevant subgroups of Prochlorococcus?

 

Sensitivity (detection limit)

 

 

 

Modelling

*Berube et al. eLife 2019;8:e41043.

3. Resolution: Prochlorococcus ASVs

3. Resolution: Prochlorococcus ASVs

3. Resolution: Prochlorococcus ASVs

3. Use cases:

 

Resolution

• Are short amplicons (~373 bp) enough to resolve ecologically-relevant subgroups of Prochlorococcus?

 

Sensitivity (detection limit)

• Can we sequence deeply enough to find rare, but biologically-relevant organisms like diazotrophs?

 

Modelling

3. Sensitivity: Diazotrophs

UCYN-A (diazotroph)

UCYN-A (diazotroph)

qPCR (nifH) data

3. Sensitivity: Diazotrophs

UCYN-A - Braarudosphaera correlation

UCYN-A (diazotroph)

3. Use cases:

 

Resolution

• Are short amplicons (~373 bp) enough to resolve ecologically-relevant subgroups of Prochlorococcus?

 

Sensitivity (detection limit)

• Can we sequence deeply enough to find rare, but biologically-relevant organisms like diazotrophs?

 

Modelling

• How can data be used to improve modelling efforts?

3. Modelling: Rationale

• Microorganisms process up to 50% of PP
• Not yet included in trait-based models such as DARWIN
• Amplicons can potentially constrain microorganism:
  • Biogeography
  • Depth distributions
• Use to inform and test modelling approaches?

3. Modelling: ASV (~species) depth profiles

Marine actinobacterium

Extreme oligotroph

Very small size

3. Modelling: Aggregated profiles

3. Modelling: Expanding ocean coverage

3. Modelling: Expanding ocean coverage

3. Modelling: Defining traits

3. Modelling: Integrating different data types

Conclusions: Universal Primers

• Cost-effective tool to map whole cellular community from A-Z (Archaea + Bacteria + Organelles + Protists + Zooplankton)
  • Chloroplast data more quantitative vs. 18S
• Can guide to where we want to concentrate more focused studies of same DNA (e.g. meta'omics)
• Straightforward to process and interpret
• Can provide guideposts for modelling efforts

https://www.egeotraces.org/sections/GP13_Fe_D_CONC.html

Acknowledgements

• Yi-Chun Yeh (for collaboration on amplicon analysis)
• Shengwei Hou (for help and inspiration on computational analyses)
• Clark Richards (for help with R "oce")
• Ken Youens-Clark (for help with C-microbe-MAP)
• Emily Zakem (for helping me understand modelling)
• Mike Lee (for help separating 16S and 18S in silico)
• Fuhrman lab crew
  • fuhrman-lab.github.io
  • tinyurl.com/MGPrimerEval

% Eukaryotic amplicons

• GP13 (mostly oligotrophic surface waters)
  • AVG = 3.2% (range: ~1-28%)
  • For 200,000 sequences = 6400 18S sequences (2000 - 56,000)
• GA03 (mostly oligotrophic, more deep-sea sampling)
  • AVG = 2.7% (range: 0.17-17%)

DADA2 vs deblur (ASVs vs. MG)

DADA2 vs deblur (ASV vs. ASV)