MetaQuantome

An integrated, quantitative metaproteomics tool reveals connections between taxa, function and protein expression in microbiomes.

Caleb Easterly, Ray Sajulga, Subina Mehta, Praveen Kumar, Shane Hubler, Bart Mesuere, Joel Rudney,

Timothy Griffin, Pratik Jagtap

University of Minnesota, Minneapolis, MN; Rhapsody Data LLC, Madison, WI; Ghent University, Ghent, Belgium

Multiple studies have shown correlation of microbial composition with physiological conditions.

Metaproteomics approach goes beyond studies of compositional dynamics and sheds light on mechanistic details of microbial interactions with host / environment.

Metagenomics: DNA Sequencing identifies species present within complex community (16S rRNA and Whole Genome Sequencing).

Metatranscriptomics: RNA Sequencing identifies species present and possible functions within complex communities (RNASeq).

Metaproteomics: The large-scale characterization of the entire protein complement of environmental microbiota at a given point in time. Potential to unravel the mechanistic details of microbial interactions with host / environment by analyzing the functional dynamics of the microbiome.

Microbiome: Microbial genetic potential and response

Image from https://thedoctorweighsin.com/what-everyone-should-know-about-the-infant-microbiome/

DATABASESEARCH

&

STRATEGIES

•

•

DATABASEGENERATION

FASTQ

Protein / Peptide FASTA

TAXONOMYANALYSIS

Unique Peptides

FUNCTIONALANALYSIS

Proteins

Known Function

Peptides

Search Algorithm

Spectra

QUANTITATIVEANALYSIS

Spectral counts OR

Intensity data

Hypothetical Function

Unknown Function

Shared TaxonomyUnassigned Taxonomy

Metaproteomics Workflow

Metaproteomics Quantitation

Metaproteomics Quantitation

C

D D

CB

A

Benchmarking Datasets

Used a mock taxonomy dataset and the UPS dataset as benchmark datasets

Taxonomy Function

Method # Terms MSE # Terms MSE

summarization 31 0.92 713 26.2

metaQuantome 37 0.36 1718 24.4

In both the taxonomic and functional benchmarking analyses, metaQuantome quantified

more terms with higher accuracy than standard summarization-based methods.

MSE=mean squared error

Mock Taxonomy Dataset DOI: 10.1038/s41467-017-01544-xPRIDE Project PXD006118

UPS1 and UPS2 datasetsPRIDE Project PXD000279

Case Study :

Sucrose-induced oral dysbiosis

• Mass spectral data was acquired from plaque samples from twelve subjects at high risk for dental caries grown in biofilm reactor in the presence (With Sucrose, or WS) and absence of sucrose (No Sucrose, or NS) (12 in each group, 24 total samples)

• Mass spectra were searched against the Human Oral Microbiome database (HOMD) to identify microbial peptides.

• Quantitation, functional annotation, and taxonomic assignment was performed in Galaxy; metaQuantome was used to analyze the results.

Rudney et al., BMC Microbiome DOI: 10.1186/s40168-015-0136-z

Abundant taxa

Oral dysbiosis results: taxonomy

Most abundant Genera in NS Most abundant Genera in WS

FUNCTION

Oral dysbiosis results: volcano plots

TAXONOMY

Oral dysbiosis results: pca plots

FUNCTIONTAXONOMY

Oral dysbiosis results: Heatmaps

FUNCTIONTAXONOMY

Taxonomy units contribution to carbohydrate metabolism

Oral dysbiosis results: Function-Taxonomy

Taxon

Pro

po

rtio

n o

f p

ep

tid

e in

ten

sity

NS WS

Taxon

Functional distribution of Streptococcaceae peptides

Oral dysbiosis results: Function-TaxonomyP

rop

ort

ion

of

pe

pti

de

inte

nsi

ty

NS WS

Functional TermsFunctional Terms

Future Directions

● Analyze more datasets (clinical and environmental)

● Alternative tools for quantitation, taxonomy and function.

● Integrate the metaproteomics workflow with an existing

metatranscriptomics quantitative analysis and

visualization workflow (ASaiM) within Galaxy.

● Investigate peptides/proteins of unknown function/taxonomy

Differential expression analysis: proteins of known (L)

and unknown (R) function

Conclusions

• metaQuantome allows for robust quantitative functional and taxonomic analysis from proteomics datasets.

o Quantitative: Supports analysis of multiple samples, including comparison across multiple experimental conditions

o Peptide-centric: Analyze taxonomy and function while mitigating the protein inference problem

o Support for function-taxonomy interaction analysis: Leverages taxonomic and functional information of the same dataset

o Flexible & Accessible: Free and open source – available on Github, Python Package Index, Bioconda, and Galaxy

galaxyp.org

Minnesota Supercomputing InstituteJames JohnsonThomas McGowanMichael Milligan

Ira CookeMelbourne , Australia

University of Minnesota

Timothy GriffinPraveen KumarCandace GuerreroSubina MehtaAdrian Hegeman (Co-I)Art EschenlauerShane HublerRay SajulgaCaleb EasterlyAndrew Rajczewski

Biologists / collaboratorsLaurie ParkerJoel RudneyManeesh BhargavaAmy SkubitzChris WendtBrian CrookerSteven FriedenbergKevin VikenKristin BoylanMarnie PetersonSomiah AfiuniBrian SandriAlexa PragmanWanda WeberAmy Treeful

Harald Barsnes Marc Vaudel University of Bergen, Norway

University of Freiburg,Freiburg, Germany

VIB, UGhent, Belgium

Judson HerveyNaval Research InstituteWashington, D.C.

Matt ChambersNashville, TN

Alessandro TancaPorto Conte Ricerche, Italy

Carolin KolmederUniversity of Helsinki, Finland

Thilo MuthBernhard RenardRobert Koch Institut

Thomas DoakJeremy Fisher Indiana University

Josh EliasStanford University

Brook NunnU of Washington

Lennart Martens (Co-I)Bart MesuereRobbert G Singh

Bjoern GrueningBérénice Batut

Lloyd Smith (Co-I)Michael ShortreedUW-Madison

Karen ReddyMo HeydarianJohns Hopkins University

Anamika KrishanpalPriyabrata PanigrahiPersistent Systems Limited

Stephan KangIntero Life Sciences

galaxyp.org

FundingACKNOWLEDGMENTS

Magnus Øverlie Arntzen (Co-I)Francesco DeloguNMBU,Oslo, Norway