An Expert System for Scoring DNA Database Profiles Dr. Mark W. Perlin Cybergenetics Pittsburgh, PA.

Cybergenetics

An Expert System for ScoringDNA Database Profiles

Dr. Mark W. PerlinDr. Mark W. PerlinCybergeneticsCybergeneticsPittsburgh, PAPittsburgh, PA

Cybergenetics

Outline

• • STR Forensic DatabasesSTR Forensic Databases• • The Data Scoring ProblemThe Data Scoring Problem• • An Expert System SolutionAn Expert System Solution• • Complex DNA AnalysisComplex DNA Analysis

Cybergenetics

Generate STR Profiles

STR lengthsSTR lengths

PCR productsPCR products

PCR amplificationPCR amplification

electrophoretic bandselectrophoretic bands

size separationsize separation

data pixelsdata pixels

data acquisitiondata acquisition

• • PCR stutterPCR stutter• • preferential amppreferential amp• • contaminationcontamination

• • peak spreadpeak spread• • lane crosstalklane crosstalk• • size variationsize variation

• • baseline shiftbaseline shift• • dye bleedthroughdye bleedthrough• • size distortionsize distortion

Data artifactsData artifacts

Cybergenetics

Analyze STR Profiles

STR lengthsSTR lengths

PCR productsPCR products

electrophoretic bandselectrophoretic bands

data pixelsdata pixels

designate allelesdesignate alleles

size & quantitatesize & quantitate

recover DNA signalsrecover DNA signals

check allele callscheck allele callsfor PCR errorsfor PCR errors

examine signalsexamine signalsfor separation errorsfor separation errors

inspect datainspect datafor acquisition errorsfor acquisition errors

Human data editingHuman data editingQuality AssuranceQuality Assurance

Cybergenetics

High-Quality STR Data ScoringGenerate STR DataGenerate STR Data

Forensic DNA DatabaseForensic DNA Database

Cybergenetics

Data Analysis: Labor Cost≥ ≥ $1.00 per experiment$1.00 per experiment

Cybergenetics

Computer Data Scoring

STR lengthsSTR lengths

PCR productsPCR products

electrophoretic bandselectrophoretic bands

data pixelsdata pixels

designate allelesdesignate alleles

size & quantitatesize & quantitate

recover DNA signalsrecover DNA signals

check allele callscheck allele callsfor PCR errorsfor PCR errors

examine signalsexamine signalsfor separation errorsfor separation errors

inspect datainspect datafor acquisition errorsfor acquisition errors

Expert SystemExpert SystemQuality AssuranceQuality Assurance

fast, accurate, objectivefast, accurate, objective

Cybergenetics

TrueAllele™ AutomationFlexible automated analysis:Flexible automated analysis:

DNA fragment sizing and quantitationDNA fragment sizing and quantitation

Any gel or capillary sequencer dataAny gel or capillary sequencer data

image analysissignal analysis

allele assignmentquality assessment

Automated Analysis:• • MacintoshMacintosh• • WindowsWindows• • UNIXUNIX

Focused user review of only the 5%-10% suspect dataFocused user review of only the 5%-10% suspect data

Cybergenetics

(1) Input

Cybergenetics

Jacob 1• acquire data• process signal• separate colors• remove primers• track sizes• extract profiles

(2) Run Processing & Q/A

Cybergenetics

Cybergenetics

Hitachi Image Data

Cybergenetics

ABI/3700

Cybergenetics

MegaBACE

Cybergenetics

Jacob 2

• derive allelic ladder• transform coordinates• quantitate trace• call alleles

(3) Allelic Processing & Q/A

Cybergenetics

AlleleView

Cybergenetics

(4) Output

Cybergenetics

Quality AssuranceUpdate

Test

Assemblecross-platform

software

AssemblePDF hypertextuser manuals

E-distribution

Cybergenetics

www

www.cybgen.comwww.cybgen.comwww.cybgen.com

Cybergenetics

Automated Scoring: FSS/UKGenerate STR DataGenerate STR Data

UK National DNA DatabaseUK National DNA Database

Human reviewsHuman reviewsjust 5%-10%just 5%-10%of the dataof the data

TrueAllele expert systemTrueAllele expert systemscores all STR data and scores all STR data and assesses data qualityassesses data quality

Cybergenetics

NIJ Validation Project

Automatically rescore 30,000 samplesAutomatically rescore 30,000 samples

Florida database Florida database • ABI/310               • ABI/310                • ABI/3700 • ABI/3700Virginia database Virginia database • Hitachi/FMbio2• Hitachi/FMbio2  

Compare with CODIS profilesCompare with CODIS profiles

Cybergenetics

Complex DNA Analysis

• • Remove PCR stutterRemove PCR stutter• • Adjust relative amplificationAdjust relative amplification• • Quantitate DNA band overlapQuantitate DNA band overlap

Automatically resolve data artifactsAutomatically resolve data artifacts

Mathematical modelMathematical model

• • Resolve DNA mixturesResolve DNA mixtures

Cybergenetics

Mixture Deconvolution

a+ba+b

aa

bb

0.30.3

0.70.7

––

Cybergenetics

Automated DNA Profiling

STR Data PathwaySTR Data Pathway

• Prepare SamplePrepare Sample• Amplify DNAAmplify DNA• Separate SizesSeparate Sizes

• Match DatabaseMatch Database• Designate AllelesDesignate Alleles