Post on 08-Mar-2018
The Future of PredictiveThe Future of PredictiveToxicology in a Flattening World:Toxicology in a Flattening World:Barriers to greater validation and acceptanceBarriers to greater validation and acceptanceof QSAR models in regulatory applicationsof QSAR models in regulatory applications
Ann RichardAnn RichardNational Center for Computational ToxicologyNational Center for Computational Toxicology
US Environmental Protection AgencyUS Environmental Protection Agency
Toxicity Risk Assessment
NO2 increasing complexity
increasing relevance to RA
increasing uncertainty
SARSARstructure-activitystructure-activity
relationshipsrelationships
ToxicityToxicityPredictionPredictionProblemProblem
A (Q)SAR model for regulatory purposes shouldA (Q)SAR model for regulatory purposes shouldbe associated with the following information:be associated with the following information:
•• a defined endpointa defined endpoint
•• an unambiguous algorithm, i.e. transparency an unambiguous algorithm, i.e. transparency
•• a defined domain of applicability a defined domain of applicability
•• appropriate measures of goodness-of-fit, robustness, and appropriate measures of goodness-of-fit, robustness, andpredictivitypredictivity
•• a mechanistic interpretation, if possible a mechanistic interpretation, if possible
OECD Principles OECD Principles ((SetubalSetubal, 2002), 2002)::
A (Q)SAR prediction for regulatory purposes should be placed in alarger context of information to assess the utility of the prediction.
EPAEPA’’s Revised Cancer Risks Revised Cancer RiskAssessment GuidelinesAssessment Guidelines
Mode of Action: Mode of Action: ““Use of mode of action information in hazardUse of mode of action information in hazardcharacterization ... is a central part of the proposed approach ...characterization ... is a central part of the proposed approach ...””
Structural Analogue Data:Structural Analogue Data: ““Confidence in conclusions is a functionConfidence in conclusions is a functionof how similar the analogues are to the agent under review inof how similar the analogues are to the agent under review instructure, metabolism, and biological activity.structure, metabolism, and biological activity.””
Structural-Activity Relationships: Structural-Activity Relationships: ““SAR analyses and models canSAR analyses and models canbe used to predict molecular properties, surrogate biologicalbe used to predict molecular properties, surrogate biologicalendpoints, and carcinogenicity. ... Suitable SAR analysis [ofendpoints, and carcinogenicity. ... Suitable SAR analysis [ofchemicals that do not bind covalently to DNA] requires knowledge orchemicals that do not bind covalently to DNA] requires knowledge orpostulation of the probable mode(s) of action ...postulation of the probable mode(s) of action ...””
Narrative: Narrative: ““biological plausibilitybiological plausibility””, , ““mode of action consistent withmode of action consistent withgenerally agreed-upon principles and understanding ofgenerally agreed-upon principles and understanding ofcarcinogenicitycarcinogenicity””
Validation: Validation:What is the performance of model outside of the trainingWhat is the performance of model outside of the trainingset? in different chemical domains?set? in different chemical domains?
Domain(sDomain(s) of applicability:) of applicability:When should a model be applied?When should a model be applied?When not?When not?
Confidence level: Confidence level:What confidence can one place in the prediction?What confidence can one place in the prediction?
Important SAR Modeling Concepts:Important SAR Modeling Concepts:
Knowledge includes knowing what you don’t knowKnowledge includes knowing what you don’t know
Toxicity prediction
Global SAR Toxicity Prediction Model
Q
Chemical Structures
Descriptors
Activities
low
high
Confidence inprediction
Chemical Structures
Descriptors
Activities
Activity predictionConsistent with known mechanism of actionBiological & chemical plausibilityModel performance for analoguesIdentified analoguesRelevant descriptorsCoverageStatistical certainty
low
high
Confidence inprediction
Q
Global SAR Toxicity Prediction Model
Model assumptionsTraining databaseData qualityStatistical performance measuresProspective predictivityModel failuresModel complexityInterpretability
Example: Example: DEMETRA ProjectDEMETRA ProjectHybrid System for prediction of aquatic toxicityHybrid System for prediction of aquatic toxicity
(combination of 5 Neural Network models)(combination of 5 Neural Network models)
-3
-1
1
3
5
7
-3 -1 1 3 5 7
Observed
Predicted
TRAINING SETTRAINING SET
y = 0.90x + 0.26y = 0.90x + 0.26
RR22 = 0.76 = 0.76
TEST SET
y = x + 0.52
R2 = 0.72
What sorts of chemicals areWhat sorts of chemicals are well predicted? well predicted? poorly predicted? poorly predicted?
For a test chemical prediction:For a test chemical prediction:
What are closest model analogs? What are closest model analogs?
How accurate are predictions for How accurate are predictions forthe closest the closest modelmodel analogs?analogs?
How well are the closest How well are the closest chemicalchemicalanalogs predicted?analogs predicted?
Are descriptors well within range Are descriptors well within rangeof dataset?of dataset?
Is prediction plausible? Is prediction plausible?
Draize rabbit eye test scores: 68 Bulk liquids – direct eye exposure Modified maximum average scores (MMAS)
Human eye irritation thresholds: 23 compounds - vapor phase exposure Eye Irritation Threshold (EIT)
Adjusted by liquid-saturated vapor pressure, PAdjusted by liquid-saturated vapor pressure, P0 0 ::
Log (1/EIT) = log (MMAS/ PLog (1/EIT) = log (MMAS/ P0 0 ))
Single QSAR for combined set of 91Single QSAR for combined set of 91compoundscompounds
::5 solubility-related parameters, R5 solubility-related parameters, R22=0.936=0.936
Supports relevance ofDraize test to humans
Mechanistic basis ofQSAR: passive transfer
Applicable to low vpchemicals
Applicable tohazardous chemicals
Extends range of bothdatasets
Organophosphate
Combining SAR and Biofunctional InformationPredicting Carcinogenicity of Organophosphates
In vitrogenotoxicity?
+
-
- High acutetoxicity?
MarRapidhydrolysis? +
-M
...
+
-
Increase serumalkalinephosphatase?
LM
- HM
SAR:Structuresuggestiveof alkylation?
Strong chelating property?
LM
??
+
+
??
????
In vivogenotoxicity?
L
SAR
SAR SAR
SAR
Woo et al. (1996) Environ. Woo et al. (1996) Environ. CarcCarc. & . & EcotoxEcotox. Revs., C14:1-42. Revs., C14:1-42
CHEMICAL XPrioritized for weight of evidence
Determination of Genotoxic Carcinogenicity
Available Empirical DataOn Carcinogenicity
Available Eimpirical Dataon Genotoxicity
QSAR Model Predictions For Carcinogenicity
SAR Model PredictionsFor Carcinogenicity
QSAR Model Predictionsfor Genotoxicity
SAR Model Predictionsfor Genotoxicity
WEIGHT OF EVIDENCECONCLUSION FORCARCINOGENICITY
WEIGHT OF EVIDENCECONCLUSION FOR
GENOTOXICITY
Model RobustnessIndicators
Model RobustnessIndicators
Sufficient Weight of Evidence for/against
Carcinogenicity?
WEIGHT OF EVIDENCECONCLUSION FOR
GENOTOXIC CARCINOGENICITY
NO
NO
NO
NO
YES
YES
YES
YES
Sufficient Weight of Evidence –
Positive or Negative - for
Carcinogenicity?
Sufficient Weight of Evidence for/against
Genotoxicity ?
Sufficient Weight of Evidence –
Positive or Negative - for Genotoxicity ?
Coutesy of Bette Meek, Health Canada
ComHaz Preliminary Weight of Evidence – Data/(Q)SAR
SAR
SAR
Integrated Testing Scheme:Integrated Testing Scheme:New Chemical Hazard AssessmentNew Chemical Hazard Assessment
80% prediction accuracy:Propose use with reduced
confirmatory assay
Pilot Study (10 chem)Pilot Study (10 chem)In vivo results availableIn vivo results available
(use DEREK, TOPKAT, misc. SAR models)(use DEREK, TOPKAT, misc. SAR models)
Implementationwould lead to 37%reduction in use
of animals
Gubbels-van Hal et al, Regul. Tox. Pharm., 2005, 42:284-295.
PhyschemPhyschem Properties PropertiesMP, BP, MP, BP, solubsolub, , LogPLogP, VP, .., VP, ..
Tox PropertiesTox Properties acute oral acute oral subacutesubacute toxtox (28 day) (28 day) acute dermal/inhalation acute dermal/inhalation eye/skin irritation eye/skin irritation skin sensitization skin sensitizationin vitro in vitro mutagmutag (2 tests) (2 tests)
EcotoxEcotox Properties Properties biodegbiodeg, acute daphnia, , acute daphnia, …… acute fish acute fish toxtox
80-90% predictionaccuracy:
Propose use with singleconfirmatory assay
Poor prediction accuracy (10%)No reduction in animal use
Develop newSAR model
Structure-based Screening & Structure-based Screening & Prioritization:Prioritization:
Chemical(s)of concern
Apply existingSAR model
DataData
DataData
DataData
DataData
DataData
DataData
Place in context ofexisting data andunderstanding
Mine existingdata forstructural &biofunctionalanalogs
Chemistry-based Data Mining Chemistry-based Data Mining & Exploration:& Exploration:
Chemical(s)Chemical(s)of concernof concern
StructuralStructuralanalogsanalogs
Chemical-Chemical-specificspecific
datadata
PropertyPropertyanalogsanalogs
Biological/Biological/mechanisticmechanistic
analogsanalogs
Structure-Activity Relationships
Individualexperiment results
Cannot download entire list of NTP Cannot download entire list of NTPchemicals and test summary datachemicals and test summary data
Cannot structure or substructure-search Cannot structure or substructure-searchdatabasedatabase
Cannot download subsets of data: Cannot download subsets of data: list of TA98 pos data list of TA98 pos data
list of all thyroid tumor carcinogens list of all thyroid tumor carcinogens
Cannot ask relational questions of data: Cannot ask relational questions of data: what chemicals are TA100 what chemicals are TA100 negneg + TA98 pos? + TA98 pos?
list all chemicals with positive rat liver tumor list all chemicals with positive rat liver tumorfindings in cancer bioassay that are also non-findings in cancer bioassay that are also non-mutagenicmutagenic
Short-TermBioassays
CAS, Chemical Name
Developmental Studies
Embryo/fetalmalformations
NTP Database Site Map
““Shared standards are a huge flattener,Shared standards are a huge flattener,because they both force and empower morebecause they both force and empower morepeople to communicate and innovate overpeople to communicate and innovate overmuch wider platforms.much wider platforms.””
The World is FlatThe World is Flat - Thomas Friedman, 2004. - Thomas Friedman, 2004.
Gene expression
BioactivityProtein expression
Cell-basedassays
Chronicwholeanimalstudies
Chemical Structures
Neurotox
Repro &Develop Tox
Cancer
Genetox
DDistributedistributed
SStructuretructure--SSearchableearchable
ToxToxicityicityPublicPublicDatabaseDatabaseNetworkNetwork
DDistributedistributed
SStructuretructure--SSearchableearchable
ToxToxicityicityPublicPublicDatabaseDatabaseNetworkNetwork
Chemical structure-annotation
Data standards and integration
-nature & CAS of mixturecomponents,-tautomers,-sterechemistry,-error in source info-replicate 2D, CAS, parent-quaternary ammonium
STRUCTURE_ChemicalType
STRUCTURE STRUCTURE_Shown TestSubstance_ChemicalName
STRUCTURE_TestedForm_DefinedOrganic
STRUCTURE_MolecularWeight
STRUCTURE_Formula
STRUCTURE_IUPAC
STRUCTURE_SMILES
STRUCTURE_Parent_SMILES
STRUCTURE_InChI
ChemicalNotedefined organicinorganicorganometallic
parent,salt Na, Cl, etccomplex HCl, H2O, mesylate, etc
tested chemical,general form of chemical,active ingredient of formulation,representative isomer inmixture,representative component inmixture,monomer of polymer, simplified to parent
CH3
O
CH3
DSSTox Standard Chemical Fields:DSSTox Standard Chemical Fields:
TestSubstance_CASRN
TestSubstance_Description
single chemical compoundmacromoleculedefined mixture or formulationundefined mixtureunspecified or multiple forms
DSSTox_SID
NCTRlogRBA
ER RBA
ChemClass ERB
Activity GroupERB
RationaleChemClass ERB
MeanChemClass ERB RBA
LogP
F1, F2, …F6
ChemClass FHM
MOA
MOACONF
CLOGP
LC50
LC50NOTE
LC50RATIO
MIXMOA
TOXINDEX
FATS
BEHAVIOR
ChemClass DBP
Concern Level
Rationale
Rational Source
AnalogChemName
AnalogCAS
AnalogSMILES
SAL CPDB
TD50 Rat
TD50Mouse
Target Sites RatMale
Target SitesRat Female
Target SitesMouse Male….Other Species
Integrating Diverse Databases from aIntegrating Diverse Databases from aChemical Structure Perspective:Chemical Structure Perspective:
CPDBCPDB DBPCAN EPAFHM NCTRER DBPCAN EPAFHM NCTRER ……..
NCTRlogRBA
ER RBA
ChemClass ERB
Activity GroupERB
RationaleChemClass ERB
MeanChemClass ERB RBA
LogP
F1, F2, …F6
ChemClass FHM
MOA
MOACONF
CLOGP
LC50
LC50NOTE
LC50RATIO
MIXMOA
TOXINDEX
FATS
BEHAVIOR
ChemClass DBP
Concern Level
Rationale
Rational Source
AnalogChemName
AnalogCAS
AnalogSMILES
SAL CPDB
TD50 Rat
TD50Mouse
Target Sites RatMale
Target SitesRat Female
Target SitesMouse Male….Other Species
Integrating Diverse Databases from aIntegrating Diverse Databases from aChemical Structure Perspective:Chemical Structure Perspective:
CPDBCPDB DBPCAN EPAFHM NCTRER DBPCAN EPAFHM NCTRER ……..
Standard Chemical FieldsStandard Chemical Fields
Standard Tox Fields: Standard Tox Fields: StudyType, Species, Endpoint, RouteStudyType, Species, Endpoint, Route
Activity Fields: Activity Fields: Dose, ActivityCategory, Dose, ActivityCategory, SummaryCallSummaryCall, ..., ...
DSSTox Database Design:DSSTox Database Design:
ToxicologyToxicology ChemistryChemistry
Toxicological Data Chemical
Structures &Properties
ContextContext Utility for SARUtility for SAR
RelevanceRelevance
Toxicologists
Biologists
Risk Assessors
Domain Expertise
SAR Modeling
Machine Learning
Comp. Chemistry
3D QSAR
CPDBAS_v3a_1481_22Oct2005
STRUCTUREDSSTox_SIDDSSTox_ID_FileNameSTRUCTURE_FormulaSTRUCTURE_MolecularWeightSTRUCTURE_ChemicalTypeSTRUCTURE_TestedForm
_DefinedOrganicSTRUCTURE_ShownTestSubstance_ChemicalNameTestSubstance_CASRNTestSubstance_DescriptionChemicalNoteChemicalReplicateCountSTRUCTURE_ChemicalName
_IUPACSTRUCTURE_SMILESSTRUCTURE_Parent_SMILESSTRUCTURE_InChIStudyTypeEndpointSpecies
SAL_CPDBTD50_Rat_mg/kg/dayTD50_Rat_mmol/kg/dayTargetSites_Rat_Male, Female, Both SexesTD50_Mouse_mg/kg/dayTD50_Mouse_mmol/kg/dayTargetSites_Mouse_Male, Female, Both SexesTD50_Hamster_mg/kg/dayTD50_Hamster_mmol/kg/dayTargetSites_Hamster_Male, Female, Both SexesTD50_Dog_mg/kg/dayTargetSites_DogTD50_Rhesus_mg/kg/dayTargetSites_RhesusTD50_Cynomolgus_mg/kg/dayTargetSites_CynomolgusActivityCategory_SingleCellCallActivityCategory_MultiCellCallNTP_TechnicalReportToxicityNote
adr = adrenal gland;bon = bone;cli = clitoral gland;eso = esophagus;ezy = ear/Zymbal’s gland;gal = gall bladder;hag = harderian gland;hmo = hematopoietic system;kid = kidney;lgi = large intestine;liv = liver;lun = lung;meo = mesovarium;mgl = mammary gland;mix = mixture;myc = myocardium;nas = nasal cavitynrv = nervous system;orc = oral cavityova = ovary;pan = pancreas;per = peritoneal cavity;pit = pituitary gland;pre = preputial gland;pro = prostate;ski = skin;smi = small intestine;spl = spleen;sto = stomach;sub = subcutaneous tissue;tba = all tumor bearing animals;tes = testes;thy = thyroid gland;ubl = urinary bladder;ute = uterus;vag = vagina;vsc = vascular system.
NTPGTZ_v1a_1921SAL_TA100, rat S9, hamster S9, w/o S9, sumSAL_TA102, …SAL_TA104SAL_TA1535SAL_TA1537SAL_TA97SAL_TA98
IMMTOX:IMMTOX:Immunotoxicity TestImmunotoxicity TestBatteryBattery
88 chemicals88 chemicals
18 immunotox measures18 immunotox measures
Summary calls Summary calls
Usage categories Usage categories AntibodyResponse AntibodyResponse NaturalKillerCells NaturalKillerCells LymphocyteProliferation LymphocyteProliferation MixedLeukocyteResponse MixedLeukocyteResponse LeukocyteCountLeukocyteCount ThymusBodyWt ThymusBodyWt SpleenBodyWt SpleenBodyWt Lipopolysaccaride Lipopolysaccaride DelayedTypeHypersensitivity DelayedTypeHypersensitivity CytotoxicTLymphocyte CytotoxicTLymphocyte SurfaceMarkers SurfaceMarkers ContactSensitizationContactSensitization HostResistanceAssaysHostResistanceAssays (6) (6)
Up next ...Up next ... IMMTOXIMMTOX: Immunotox Battery Database : Immunotox Battery Database Expanded from 1992 publicationExpanded from 1992 publication
reporting a battery of immunotox results for 88 chemicals, most from the NTP.reporting a battery of immunotox results for 88 chemicals, most from the NTP.
NCTRARNCTRAR: FDA: FDA’’s National Center for Toxicological Research - Androgens National Center for Toxicological Research - AndrogenReceptor Binding Database Receptor Binding Database Androgen receptor relative binding affinities tested inAndrogen receptor relative binding affinities tested ina common in vitro assay for 202 chemicals, provided with chemical class-baseda common in vitro assay for 202 chemicals, provided with chemical class-basedstructure activity features.structure activity features.
NTPGTZNTPGTZ: NIEHS National Toxicology Program Gene-Tox Database (E.: NIEHS National Toxicology Program Gene-Tox Database (E.Zeiger) Zeiger) Battery of genetic toxicity test results for over 1900 chemicals from historicalBattery of genetic toxicity test results for over 1900 chemicals from historicalNTP studies.NTP studies.
UNLVSSUNLVSS: UniLever: UniLever’’s Skin Sensitization Databases Skin Sensitization Database Skin sensitization resultsSkin sensitization resultsfor over 200 chemicals from Unilever studies.for over 200 chemicals from Unilever studies.
Structure-Index Files:
IRISSIIRISSI: EPA: EPA’’s Integrated Risk Information System (IRIS)s Integrated Risk Information System (IRIS)
EPAHPVEPAHPV: EPA: EPA’’s High Production Volume Chemicalss High Production Volume Chemicals
NTPTSINTPTSI: National Toxicology Program: National Toxicology Program’’s Test Resultss Test Results
DSSTox ChemoinformaticsDSSTox ChemoinformaticsTotal Records: 6625Total Records: 6625Total Unique Records: 3967 (no replicates)Total Unique Records: 3967 (no replicates)
0
500
1000
1500
2000
2500
CPDBAS
DBPCAN
EPAFHM
NCTRER
FDAMDD
NTPIMT
NCTRAR
ULVSSDIRISSI
NTPGTZ
Duplicate
Unique
DSSTox Coordination/Collaborations:DSSTox Coordination/Collaborations:
DSSTox
Lhasa:VITIC
PubChem
NCI Govt.Databases
StructureStructurerelationalrelationalsearchingsearching
ToxML
NTP
Tox DataTox DataStandardsStandards
CEBS
ArrayExpressArrayExpress,,GEO, CTD, GEO, CTD, ……
StandardStandardChemicalChemicalFieldsFields
LeadScopeSimulation PlusBioRadACD/Labs
Stanford MachineLearning Datasets
Data FilesData Files
ModelsModels
Toxicity Experimental Data Toxicity Experimental Data Summary Data: Summary Data:
ToxML
++ _ _
Activity categories Activity categories
Potency categories Potency categories
Mode of action categories Mode of action categories
Summary callsSummary calls
DSSToxDSSToxStandardStandardChemicalChemical&Tox Fields&Tox Fields
Toxicity Experimental Data Toxicity Experimental Data Summary Data: Summary Data:
ToxML
++ _ _
IntermediateIntermediatetoxicitytoxicity
classificationsclassificationsfor SARfor SAR
Activity categories Activity categories
Potency categories Potency categories
Mode of action categories Mode of action categories
Summary callsSummary calls
DSSToxDSSToxStandardStandardChemicalChemical&Tox Fields&Tox Fields
http://pubchem.ncbi.nlm.nih.gov/search/
850,000 Substances from “Legacy” Public Sources 650,000 Unique Structures 10,000,000 BioAssay Results from NCI / DTP
structure/sub-structure searching links to protein structures links to PubMed bioactivity screens similar activity profiles bioassay data
InChI=1.0/C9H6O6/c10-7(11)4-1-2-5(8(12)13)6(3-4)9(14)15/h1-3H,(H,10,11)(H,12,13)(H,14,15)
Unique Unique
Public Public
Text XML-based Text XML-based
Chemically robust Chemically robust charge state charge state chiralchiral centers centers tautomeric form tautomeric form
InChI:InChI:
PubChem (>3M) NCI (>23 M) KEGG (>10K) NIST Chemical Web Book Berkeley Carcinogenic
Potency Project DSSTox EPA Substance Registry
System
World-Wide WebWorld-Wide Web
““ElectronificationElectronification”” of historical data of historical data
Content annotation of unstructured data Content annotation of unstructured data
Chemical structure-annotation Chemical structure-annotation
Data standardization & integration Data standardization & integration
EPA Data Challenges:
InChI text annotation
EPA Structure-browser
Collaborations with NCI,ATSDR, ECB, FDA, NIEHS
Government-wide dataGovernment-wide data
EPA-wide dataEPA-wide data
CBICBIdatadata DSSTox
NIEHS/National Center for Toxicogenomics:NIEHS/National Center for Toxicogenomics:Chemical Effects in Biological System Knowledge-Chemical Effects in Biological System Knowledge-Base Base (M. Waters and J. (M. Waters and J. FostelFostel))
Metabonomics
Historical NTPToxicity data
Geneexpression
GeneFunction
Proteomics
GenePathways
DSSTox / CEBS Collaboration:DSSTox / CEBS Collaboration:Part 1 Part 1 –– DSSTox Annotation of CEB DSSTox Annotation of CEB
Metabonomics
HistoricalToxicity data
Geneexpression
Gene Function
Proteomics
GenePathways
Historical NTPToxicity data
DSSTox Standard Chemical Fields
Structure-Structure-searchabilitysearchability
AnalogAnalogsearchingsearching
Cross domainCross domainsearchingsearching
DSSTox / CEBS Collaboration:DSSTox / CEBS Collaboration:Part 1 Part 1 –– DSSTox Annotation of CEB DSSTox Annotation of CEB
Metabonomics
HistoricalToxicity data
Geneexpression
Gene Function
Proteomics
GenePathways
Historical NTPToxicity data
DSSTox Standard Chemical Fields
Structure-Structure-searchabilitysearchability
AnalogAnalogsearchingsearching
Cross domainCross domainsearchingsearching
DSSTox Standard Chemical Fields
DB7DB6DB5DB4 DB3 DB2 DB1
DSSTox Toxicity Data Files
DSSToxDSSToxDatabaseDatabaseNetwork +Network +
Part 2 Part 2 –– Link CEBS to DSSTox Database Network Link CEBS to DSSTox Database Network
DSSTox / CEBS Collaboration:DSSTox / CEBS Collaboration:Part 1 Part 1 –– DSSTox Annotation of CEB DSSTox Annotation of CEB
Metabonomics
HistoricalToxicity data
Geneexpression
Gene Function
Proteomics
GenePathways
Historical NTPToxicity data
Public Genomic Databases
Chemical Inventory Chemical Inventory
Chemical indexing Chemical indexing
Linkages Linkages
Predictive Toxicology
SAR Predictionsbased solely onchemicalstructures &properties:MCASETOPKATDEREK
Toxicitypredictionsbased on geneexpressionprofiles:Gene-LogicIconix
“Glocalization” of Predictive Tox Models
structu
restru
cture
toxicity
toxicity
chemistrychemistry biology
biology
Mech 1
Mech 2
Mech 3
Mech 5
ToxicityEndpoint
ChemicalStructures
Mech 4Biological attributesBiological attributesChemical reactivity Chemical reactivity
PNAS January 11, 2005 vol. 102 no. 2 261–266CHEMISTRY PHARMACOLOGY
1576 compoundsX 92 assays
Azole cluster bybiospectra similarity
Approximate “proteomic” signature
PASS (Prediction of ActivitySpectra for Substances)
Cerep: In vitro bioactivity profiles
Expanded chemical Expanded chemical ““propertiesproperties”” in inrelation to biological activityrelation to biological activity
Databases & models need to Databases & models need toextend beyond pharmaceuticals toextend beyond pharmaceuticals toenvironmental/industrial chemicalsenvironmental/industrial chemicalsspanning toxicity spacespanning toxicity space
Reference dataset of Reference dataset oftoxicity-related chemicals withtoxicity-related chemicals withstructures & bioactivity profilesstructures & bioactivity profiles
NIH/NCGC Roadmap:NIH/NCGC Roadmap:Small Molecules High-Small Molecules High-Throughput Screening InitiativeThroughput Screening Initiative
nihroadmap.nih.gov
> 200K > 200K ““small moleculessmall molecules””
> 200> 200bioactivity &bioactivity &cell-basedcell-basedassaysassays
7 dose7 dosedilutionsdilutions
Sample Sample ““toxicitytoxicity”” chemical chemicalspace:space:
NTP chemicalsNTP chemicals
EPA pesticides, inerts EPA pesticides, inerts
EPA HPV Chemicals EPA HPV Chemicals
DSSTox DSSTox
NCI/ NCI/ChemNavigatorChemNavigator
TOX
Biochemical“target” assays
Toxico-Chemoinformatics:Toxico-Chemoinformatics:Data Standardization, Integration, ExplorationData Standardization, Integration, Exploration
In Vitro assays
In Vivo wholeanimal studies
“In Silico” Predictions
Biochemical“target” assays
Toxico-Chemoinformatics:Toxico-Chemoinformatics:Data Standardization, Integration, ExplorationData Standardization, Integration, Exploration
In Vitro assays
In Vivo wholeanimal studies
“In Silico” PredictionsCalculated structures, properties
Tissue, organs
Short-term tests
Receptors, enzymes, proteins
Genomics
Chronic, acute
Cell-based assays