26
Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment Robert E. Synovec Associate Chair, Graduate Program Faculty Director, CPAC Department of Chemistry, Box 351700 University of Washington Seattle, WA 98195 Email: [email protected] CPAC Spring Meeting, Seattle WA May 3, 2010

Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

  • Upload
    others

  • View
    7

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Discovery Based Analytics for Bio-Fuels Characterization and

Food Quality Assessment

Robert E. SynovecAssociate Chair, Graduate Program

Faculty Director, CPACDepartment of Chemistry, Box 351700

University of Washington Seattle, WA 98195

Email: [email protected]

CPAC Spring Meeting, Seattle WAMay 3, 2010

Page 2: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

• Fundamental Studies of Separation Science Principles and Metrics

• Instrumentation and Sensor Development

• Data Analysis – Chemometrics – Software

• Methodology Design and Optimization

Advances in Separation Science Knowledge and Technology:

10 nm

Synovec Research Groupand Separation Technology

………from high-speed analysis of simple mixtures to the analysis of complex samples

Page 3: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Chemical Analysis Challenge:At the discovery-stage can we, with statistical confidence, findsignificant changes in chemical composition (from very large to very small) in extremely complex samples, for example, biological samples with literally 100’s to 1000’s of compounds?!

• Health Biomarker Discovery, Clinical Application• Feed Stock and Biochem Evaluation (eg., Biomass to Fuel)• Forensics• Environmental Analysis • Industrial Discovery (eg., Biotech)• Food Quality, Safety and Security

Changes between…Healthy vs. Diseased, Wild-type vs. Mutant, One growth condition vs. another growth condition, “Good” biomass vs. “Poor” biomass, Optimal yeast strain,Poor vs. Good Quality

Page 4: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Comprehensive Two-Dimensional Gas Chromatography (GC x GC)

FID

Sig

nal

15 Component Mixture:REAL-TIME separation into different chemical classes!

Column 1 (Non-Polar)–10-m x 320-μm i.d. –0.25-μm poly(5% diphenyl/ 95% dimethyl siloxane) –35°C initial, 120°C/min program, 25.5 psi H2

Column 2 (Polar)–2-m x 250-μm i.d. –0.2-μm cyanopropyl polysiloxane –100°C, 25.0 psi H2

Page 5: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Comprehensive Two-Dimensional Gas Chromatography with Time-of-Flight Mass

Spectral Detection (GC x GC - TOFMS)

• Complete mass spectra ⇒peak identification

• Fast ⇒ 500 spectra / secondPeak widths on column two ~ 50 ms

• Adds another selective dimension ⇒ 3rd - order technique, benefit by

using chemometric software

Page 6: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Ion

Cou

nts

Ion

Cou

nts

Ion

Cou

nts

Extracted Ion Chromatograms

m/z 217

m/z 128

m/z 73

m/z

Time 1

Data Cube

3rd Order Data• column 1 retention time • column 2 retention time • full mass spectrum at each point

We analyze the RAW data!….and apply Chemometrics

Page 7: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Complex “Real” SampleIn

tens

ity

100’s of peaks…Concentration range is

~ factor of 100,000Many peaks overlap!

GC x GC

Page 9: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Metabolomic analysis of fermenting and respiring yeast…..wild-type and mutants

High glucose

Ethanol

FermentingRepressed (R)

RespiringDerepressed (DR)

Metabolism:Transcription:

Low glucose

Ethanol

Collaboration with Ted Young, UW Biochemistry

Page 10: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

OO

O

O

O

H

H

H

H

D-Ribose

SiO

OSi

O

Si OSi

NO

D-Ribose Meox 4TMS

1) Oximation

2) Trimethyl-silylation

Methoximate and Trimethylsilylate:

(1) Reduce anomer formation (2) Increase volatility (3) Increase thermal stability (4) Decrease polarity

Derivatization - getting ready for GC x GC-TOFMS analysis

Metabolites:

Organic acidsAmino acidsSugars Sugar phosphates, etc

Page 11: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Typical data, WT yeast grown in glucose conditions

GC x GC –TOFMS of Repressed Yeast Cell Extract, m/z = 73,Metabolites have been derivatized: m/z = TMS group is a “selective” channel

•Over 590 peaks at this m/z alone - Complex !•Many data runs…a huge amount of data to process !

ISSUES:

Page 12: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Chemometric data analysis tools: utilize 3rd order data structure

(1) Discover sample-class distinguishing peak locations in 2D separation space

– Data reduction by a 3D Fisher ratio method

(2) Targeted peak analysis: 3D mathematical resolution, confirmedmass spectral compound identification and quantification

– PARAFAC GUI ….state-of-the-art software tools to apply powerful Linear Algebra concepts

Discovery-Based Approach: comprehensively explore the data using chemometric data reduction methods to “discover” the sample class -distinguishing compounds

From high throughput data reduction and analysis to valuable information !

Page 13: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

TCA Cycle

Glucose

Ethanol Acetyl CoA

glycolysis

Study Protein Function with Metabolomics (ΔSnf1 mutant study)

XX∆Snf1

cannot complete the shift

• Study this mutant strain at metabolome level• Wild type (R & DR)• Mutant (R & DR)

• In the absence of specific proteins (Snf1 Protein Complex) cells are unable to switch from using glucose to ethanol

~ 160 metabolites analyzedR- glucoseDR- ethanol

Page 14: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

0

0.00002

0.00004

0.00006

0.00008

0.0001

0.00012

0.5 2 4 6Time (hours)

Nor

mal

ized

(TIC

) PA

RA

FAC

vol

ume

Fumarate

• TCA Cycle is active in DR conditions• Snf1 protein complex needed to make shift from R

to DR conditions

TCA Cycle

Glucose

Ethanol Acetyl CoAglycolysis

Page 15: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Metabolite wt

adr1∆

cat8∆

snf1∆

adr1∆

-ca

t8∆

• Metabolites were found that differ between wt and various mutant strains related to the diauxic shift (switch from glucose to non-fermentable carbon sources)

• Provides insight to which factors are most essential in the shift

These metabolites, specifically, highlight which strains are most similar to wt (adr1∆) and which are most different (cat8∆, adr1∆cat8∆,snf1∆) Included in this list are the TCA intermediates.

Wild type vs. mutants

Carbohydrate consumptionand metabolism… a tool to optimize ethanol production.

Page 16: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

0.0

0.5

1.0

1.5

2.0

2.5

stearic acid

citrate α-ketoglutaricacid

succinate fumarate malate

Rat

io to

wt

WTadr1∆cat8∆adr1∆cat8∆snf1∆

snf1∆metabolite and RNA ratios to wt show good agreement

TCA Cycle and Metabolite Intermediates

Page 17: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

End Products: Bio-Fuel Characterization

These tools can also be used to identify compounds that differ between various bio-fuels to provide insight into the chemical similarity of the fuels and to determine which fuels would be better “drop-ins” and/or what changes can be made to the process.

Traditional FuelBiofuel A

Biofuel B

Biofuel C

Biofuel D

Page 18: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Food Quality Assessment Study

• Collaboration with Theo Chocolates – Organic, fair trade, “bean to bar” chocolate manufacturer

• Use analytical chemistry (and chemometrics) to help monitor/improve the process

BEAN BARFood Quality Assessment

for Raw Materials

www.theochocolate.com

Page 19: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Chocolate Industry: Quality Control for Raw Materials

• Quality of raw materials is very important. Fermentation is not sufficiently controlled ….moisture damage is a major concern.

• Is there a way to distinguish high quality raw materials from poor quality prior to purchase and production?

• Are there “bio-marker” compounds, that can be readily correlated to bean quality, that are independent of bean origin?

Cacao Bean Nibs

E. M. Humston, Y. Zhang, G. F. Brabeck, A. McShea, R. E. Synovec, J. Sep. Sci., 2009, 32, 2289-2295.

Page 20: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Experimental Design• To investigate effect of moisture damage

Fisher ratio analysis to find differences between damaged and undamaged beans

• Solid Phase Micro Extraction (SPME) to concentrate head space analytes (no derivatization needed)

unmolded

m/z

Col 1

molded

m/z

Col 1

unmolded molded

SurfaceChemistryMetabolism!

Predict Damage Early?!

Page 21: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Headspace analysis – for volatile and semi-volatile compounds

• Solid Phase Microextraction (SPME)

• Headspace sampling

• GC x GC-TOFMS analysis

GC/MS

Page 22: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Heat bean in vile in water bath to 60°C

Expose SPME fiber to bean headspace

vapor

GC x GC-TOFMS

Transfer SPME fiber to instrument for injection

SPME –

adapted toCacao BeanProject

Page 23: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Differences can be Identified

UNMOLDED

MOLDED

UnmoldedMolded

Tetramethyl-pyrazine

0

1E7

2E7

3E7

4E7

5E7

6E7

1 2 3 4 5 6 7 8 9 10

Peak

Are

a

Bean Number

2,3-dimethyl-pyrazine

1 2 3 4 5 6 7 8 9 100

5.0E5

1.0E6

1.5E6

2.0E6

2.5E6

Bean Number

Others are consistently elevated in Molded Samples (pyrazine compounds have “earthy” odor quality)

Some analytes are consistently elevated in Unmolded Samples

Peak

Are

a

Acetic acid

02.0E7

4.0E76.0E78.0E7

1.0E81.2E81.4E81.6E81.8E8

1 2 3 4 5 6 7 8 9 10Bean Number

2-ethyl-1-hexanol

0

5.0E6

1.0E7

1.5E7

2.0E7

2.5E7

3.0E7

1 2 3 4 5 6 7 8 9 10Bean Number

Page 24: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

PCA for Classification…….Biomarkers Independent of Geographical Origin !

2‐methyl‐3‐Buten‐2‐oltetramethyl‐pyrazine, trimethyl‐pyrazine, 2‐nitro‐pentane, malonic acid, bis(2‐TMS ethyl ebenzenemethanol4‐cyanocyclohexenecyclopropane, 2‐(1,1‐dimethyl‐propanoic acid, 2‐methyl‐, 2,2‐2,5‐cyclohexadiene‐1,4‐dione, 2tridecane 2,6,10,14‐tetramethyl‐heptade2‐ethyl‐1‐hexanol, propanoic acid, 2‐methyl‐, 3‐hynonanalhexanoic acid

Costa RicaU

0

2

4

6

8

10

EcuadorIvory CoastM U M U M 5 10 15 20 25 300

0.1

0.2

0.3

Bean Number

Load

ings

on

PC

1 (

66.5

3%)

PCA Variables/Loadings from PARAFAC

Costa Rica

Ecuador Ivory Coast

unmolded

Col 2

m/z

Col 1

molded

Col 2

m/z

Col 1 x 3 Beans of Different Origin(U)

(M)

Page 25: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Acknowledgements

*not pictured: Ryan Wilson, Emilie Viglino, Laura Snyder,Jason Reed

Funding and Support:

CPACNIH, Honeywell-SRI-DARPA, WTC, Theo Chocolate, PNNL, CRC, ChevronTexaco

Page 26: Discovery Based Analytics for Bio-Fuels …depts.washington.edu/cpac/Activities/Meetings/documents/...Discovery Based Analytics for Bio-Fuels Characterization and Food Quality Assessment

Process Gas Chromatography with Chemometrics

W. Christopher Siegler, Jeremy S. Nadeau, Elizabeth M. Humston, Ryan B. Wilson, Jamin C. Hoggard and Robert E. Synovec

Department of ChemistryUniversity of Washington

Seattle, WA

CPAC Sponsor Meeting, Seattle WAMay 4, 2010