ToxCast Owner's Manual -
Guidance for Exploring
Data
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
About ToxCast
ToxCast is a multi-year effort launched in 2007 that uses automated chemical
screening technologies called high-throughput screening assays to expose living
cells, isolated proteins, or other biologica I molecu les to chemica Is. The cells or
proteins a re then screened for changes in biological activity that may suggest
potentia I toxic effects.
These innovative methods have the potential to limit the number of
required laboratory animal-based toxicity tests while quickly and efficiently
screening thousands of chemica Is for potential health effects.
Generating ToxCast Data
Chemicals
Assays
Resu Its P rocessi ng a nd Ana lysis
Publicly Available Data
Exploring ToxCast Data
ToxCast Data
Accessing ToxCast Data and Scenarios for Exploring Data
Citations for ToxCast Data
The figure below provides a visua I overview of the process ToxCast follows to
create an accessible database of high-throughput screening information on
thousa nds of chemica Is.
Click on the boxes in the figure below to link to more
information.
-------
EPA ^
Select, procure, and QC
chemicals
Perform high-throughput
screening as
ToxCast
Dashboard
Downloadable
Data
Explore data
-------
ToxCast Chemicals
Resources
Chemical Inventory
Chemical Landscape paper
About ToxCast
EPA's ToxCast research effort has a chemical library of approximately 4,400
unique chemicals (as of December 2017). The library includes a broad range of
chemicals with potential for human and ecosystem exposure which have
heightened regulatory concern yet are lacking data on potential health effects
of these chemicals.
The chemicals in the library have been screened in hundreds of high throughput
screening assays to generate biological activity data. The chemical library is
organized into three phases.
In 2007, Phase I was launched,screening310 chemicals (mostly pesticides)
across hundreds of assay end points. Phase II data increased the chemical
inventory to about 1,800 chemicals. In 2014, Phase III was initiated, increasing
the library to over 4,500 chemicals.
Below are links to information a bout the various ToxCast chemical lists and the
workflow EPA follows to select chemica Is, procu re chemica Is, register chemica Is,
conduct aqua lity review of the chemica Is, a nd prepa re the chemica Is for high-
throughput screening.
Links
Chemical Lists
Chemical Workflow
ToxCast Phases, Timing, and Chemical and Assay
Endpoint Counts
Phase I
Released: 2010
Phase II
Released: 2013
Phase III
Initiated: 2014
Chemical set: "phl_vl"
310 chemicals
~700 assay endpoints
Chemical set: "pbl_v2"
293 chemicals
~200 assay endpoints
Chemical set: "ph2"
768 chemicals
~900 assay endpoints
Chemical set: "elk"
799 chemicals
~50 assay endpoints
Chemical set: "ph3"
Assay testing in process
2,678 chemicals
-------
ToxCast Data Generation:
Chemical Lists
Resources
Chemical Inventory
Chemical Landscape paper
CompTox Chemistry Dashboard
About ToxCast
The list of EPA's ToxCast chemicals with publicly available high-throughput
screening data can be exported from the ToxCast Dashboard. EPA's ToxCast
Chemical lists that includes chemicals currently in process and chemicals
already screened can be downloaded from the FTP site.
FTP Download
The f i le la beled DSSTox_TOXCST_20160129.zi p conta i ns read me f i les as well as
an Excel file containing chemical information and ToxCast phase information for
each of the substances screened.
ToxCast Dashboard
Descriptions of the chemicals, their structures, and physicochemical properties
are available via the ToxCast dashboard. Users can select chemicals of interest
using a number of filters. Chemical data filters include Chemical Abstract Service
Registry Number (CASRN), chemical name, chemical category, use category,
a nd physicochemica I properties such as the octa nolwater pa rtition coefficent
(log P). For each chemical, the ToxCast Dashboard summarizes chemical
information.
Chemical Summary Charts
-------
Chemicals by Substance Type
Macromolecule,
1
Mixture of
Stereoisomers,
119 Mixture/Formulation,
159
Polymer, 11
Chemicals by Inventories of Interest
Coverage of Databases of Regulatory Interest
NHANES
GRAS
Everything Added to Food
Inventory Update Rules List
ToxRefDB
NTP Bioassay Database
IRIS Inventory
HPV
FDA Maximum Daily Dose Inventory
Carcinogenic Potency Database
61
695
2202
Numbers shown are
numbers of chemicals
tested.
953
874
193
0% 10% 20% 30%
¦ Phase 1 ¦ Phase 2 ¦ Phase 3
584
40% 50% 60%
% Database Coverage
70% 80% 90% 100%
Numbers of Tested Chemicals in Use Categories of Interest
Inert
Pharmaceutical
Pesticide
Personal Care
Fragrance
Colorant
Antimicrobial
Consumer Use
Industrial
600 800 1000
¦ Number of Chemicals
Links
Chemical Workflow
ToxCast Assays
-------
ToxCast Data Generation:
Chemical Workflow
This page describes the process EPA follows to select chemicals, procure
chemicals, register chemicals, conduct a quality review of the chemicals, and
prepare the chemicals for high-throughput screening. There is also information
about each phase of testing.
Resources
Chemical Inventory
Chemical Landscape paper
About ToxCast
On This Page
Chemical Selection and Lists
° Phase 1
Oualitv Review of Chemicals
° Phase II
Chemical Preparation Process and Timeline
° Phase III
Links
Procuring and Reeisterine Chemicals
Chemical Selection and Lists
Phase I
Chemical Types
ToxCast Phase I screened 310 unique chemicals, mostly pesticides. The
pesticides in Phase I included diverse chemical structures and a broad range of
pesticida I mechanisms. The chemica Is aIready had significa nt toxicity data
(approximately 275 compounds had nearcomplete guideline data coverage) and
a variety of chemical reactivity features and mechanistic diversity.
Pesticides were selected to be screened first because of EPA's extensive a nima I
toxicity data from testing pesticides. The ToxCast screening results from Phase I
were compa red to these animal toxicity study resu Its as proof-of-concept for
using ToxCast to evaluate chemica Is for potential health effects.
Thirty non-pesticidal environmental chemicals of research or regulatory interest
were also evaluated in Phase I, including:
-------
Perfluorinated compounds - Perfluorooctanoic acid (PFOA), a surfactant used
in fluoropolymers such as Teflon; and Perfluorooctanesulfonicacid (PFOS),
used in the semiconductor industry, ScotchGuard formulations, and flame
retardant foams.
Bisphenol A (BPA) and a set of phthalate alternatives used as plasticizers.
10 toxicologically active metabolites of included phthalatesand pesticides
such as monobutyl phthalate (a metaboliteof dibutyl phtha late) and
diazoxon (a metaboliteofdiazinon).
ToxCast Screening
The original 310 chemicals tested in ToxCast Phase I are referred to as "phl_vl".
A listing of the ToxCast phl_vl chemicals, including modified or discontinued
compounds labeled and issues annotated, is provided in the ToxCast chemical
inventory file (DSSTox_TOXCST, datelstamped 20160129) available from the EPA
DSSTox Data download page.
Toxicity Data
An extensive amount of toxicity data was already available for most Phase I
chemicals and included data for subchronic, chronic,
multigenerational reproductive, and developmental end points across several
species (rat, mouse, dog a nd ra bbit). The data was availa ble in data evaluation
records from the US EPA's Office of Pesticides Programs' review of registrant
study data. EPA researchers extracted the data from EPA's Data
Eva luation Records a nd systematically captured these data in a database of
animal toxicity data called the Toxicity Reference Database fToxRefDB).
Top of Page
Phase II
Chemical Types
Phase II retained most of Phase I and expanded the chemicals screened to
include datalpoor chemicals of research and regulatory interest. 293 unique
chemica Is from Phase I (phl_v2) were chosen, excluding chemicals which we
learned from Phase I were not suitable for high-throughput screening. These
excluded chemica Is were 14sulfurons determined to undergo rapid hydrolysis in
DMSO, 3 chemica Is deemed insufficiently soluble in DMSO,and 1 chemical
originally tested in pa rent form that was instead tested in a salt form. 768 unique
new ToxCast chemicals identified through a selection process were added in
Phase II (ph2). ToxCast Phase II also includes a list of 799 chemicals of interest to
EPA's Endocrine Disruption Screening Program (elk). Specifically, the elk library
contains many known estrogen receptor (ER) and androgen receptor (AR) active
reference chemicals. The elk list is specifically identified because the chemica Is
were not screened in the entire set of Phase II high-throughput assays, but a
specific subset for evaluation of potential endocrinelrelated activity.
The majority of Phase II chemicals were purchased from commercial suppliers
and approximately 150 chemicals were donated by research collaborators
externalto EPA. Somechemicals were donated bythechemical industry and the
U.S. Food and Drug Administration's National Center for Toxicologica I Research,
("green" plasticizer alternatives and reference liver toxicants, respectively).
Pharmaceutical companies donated the remaining 136 "failed pharma"
-------
compounds (i.e., drug candidates discontinued due to toxicity in preclinical or
clinicaltrials).Acomplete list of the donated pharmaceuticals ("donated
pharma" chemical list) is available on the ToxCast chemical libary file
(DSSTox_TOXCST, datelstamped 20160129), availablefrom the EPA DSSTox Data
download page.
Toxicity Data
The donation of failed drug compounds, along with some preclinical and clinical
data, introduced to ToxCast a set of chemicals designed to be bioactive at
specific human (or veterinary) targets. All compatible donated data on these
failed pharmaceuticals were added to EPA's Toxicity Reference data base.
ToxCast Screening
ToxCast Phase II encompassed screening of the newly added ph2 inventory in the
majority of the original Phase I ToxCast assays, as well as testing of the ph2and
reprocured phl_v2 inventory in newly acquired ToxCast and Tox21 assays. In
addition, the elk chemicals were screened in a limited subset of Phase II
endocrinelrelated assays, including many assays run using the Tox21 robotics
technology. Phase II concluded in 2015 with the public release of ToxCast assay
data for 1.863 unique ToxCast chemicals screened to that point (1,878 if
discontinued phl_vl chemicals are included in the total), including data for the
phl_vl, phl_v2, ph2, a nd elk chemica I sets.
Top of Page
Phase III
Chemical Types
ToxCast Phase III launched in late 2014. Phase III added new assays, end points,
and chemicals. Five hundred chemicals and a small set of donated mixtures and
water sa mples were added to ToxCast. Phase III also added selected chemica Is
from Tox21 and elKto test in the new assays. Unique Phase III chemicals include
fla me reta rda nts a nd chemica Is of i nterest to EPA's Endocrine Disruption
Screening Program fEDSP).
With the addition of Phase III chemicals, the total combined set of chemicals
comprising ToxCast is over 4,200 unique chemica Is. The screening and analysis
of data from Phase III chemicals is currently underway (Note: this number of
chemicals does not include chemicals only tested in Tox21).
ToxCast Screening
Phase III chemicals are not being run through all ToxCast assays. Using the
results from Phase I and Phase II, EPA identified the assays most appropriate for
the Phase III chemicals selected.
The chemical list for ToxCast Phase III is abbreviated as "ph3" chemicals and
does not overlap with earlier chemica I inventories having full or partial testing
coverage in Phases I and II (phl_vl, phl_v2, ph2, and elk). Each of these
inventories, as well as the testing phase in which the chemical was screened (I, II,
and/or III) is labeled within the ToxCast chemica I inventory file (DSSTox_-
TOXCST, date-stamped 20160129), available from the EPA DSSTox Data
download page.
-------
Top of Page
Procuring and Registering
Chemicals
EPA follows a standard process for procuring and registering chemicals which
includes purchasing chemicals from a supplierand working closely with
a contractor to register and catalogue the chemicals. The majority ofToxCast
chemicals are purchased from commercial suppliers.
1. First, EPA provides the supplier a list of generic CAS and chemical names that
need to be purchased. EPA may also provide chemical structure information
in the form of SMILES to the contractor to help with structure sea rching of
larger aggregated commercial chemical services, such as eMolecules exit
orChemNavigator
EXIT
2. EPA purchases two bottles of analytical grade samples (>98% purity) from a
supplier. A majority of sa mples a re then shipped directly from the su pplier to
the contractor in preltared, barcoded vials. One of two bottles is designated
for solubilization a nd the other stored in neatlpowder form. The chemica I
supplieralso providesa chemicalstructurefileand molecular weight for
defined pure compounds (SMILES or Structure Data Format [SDF] file).
3. When the contractor receives the chemicals, bottles a re scanned, weighed,
registered into the contractor's chemical management tracking system, and
either immediately solubilized, or stored in a I20°C freezer under inert
conditions until a solubilization order is placed. The contractor's chemical
ma nagement tracking system records the Supplier, Cata log number, Lot
number, the Contractor shipment sample code, SMILES, molecula r weight
(usua lly derived from the structu re file), physica I form of the chemica I
received (solid or liquid), qua ntity (u I or mg), a nd the date the sa mple was
registered in the system. In cases where a chemical name, CAS or structure
a re not provided, the contractor attempts to fill in this information through
reference to the origina I orders or supplier website cata logs, or through
internal database matches to the compound structures provided.
Asmall percentage of the chemicals deviate from the standard process (less
than 10% of Phase II and Tox21 chemicals). Most common deviations from this
process a re the use of specia Ity suppliers a nd procurement of la rger or sma Her
quantities for hardltollocate chemica Is.
EPA requires, wherever possible, a Certificate of Analysis (COA) and Material
Safety Data Sheet (MSDS) from the chemica I su pplier. The chemica I supplier
provides the COA and MSDS to the contractor a nd the contractor provides it to
EPA. Occasiona lly sa mples a re accepted by EPA without docu mentation when
the supplier has problems obtaining the documentation (COAs a re missing for
approximately 15% of unique lotlbatchchemicalsamples).
Top of Page
Quality Review of Chemicals
-------
The goal of chemical quality check (QC) is to have accurate and reliable
information on every chemical undergoing ToxCast screening. There are
practical limits as well as a bala nee that must be struck between
cost/time/efficiency and the larger objectives of ToxCast. In the absence of
perfect knowledge and certainty, the ToxCast chemical PC process minimizes
controllable sources of errors, particularly in the chemical information QC
review and registration process, but also in handling and storage procedures. At
the sa me time, the ToxCast a na lyticaI QC processes attempt to detect,
understand, document, and communicate actual errors and problems
impacting particular chemicals under testing conditions.
In the QC process, EPA may request its chemical management services
contractor to a na lyze chemica I sa mples to confirm the supplier's
identification of the chemical sample and to estimate sample purity (ideally
>90%for commercial grade and >99% for pharmaceutical or pestinal analytical
grade samples) using appropriate analytical methods.AnalyticalQC providesan
experimental standard of verification and is required to confirm the
chemical identity and purity in the plated DMSO solutions undergoing testing at
the ti me of plati ng, as well as at later ti me poi nts (to assess sa mple sta bility
overtime).
The QC a na lysis requires the a vaila bility of a na lytica I methods suita ble to
a na lyzi ng the va rious types of chemica I compou nds derived from the chemica I
domains identified. The standard ana lysis would consist of high-throughput,
low-resolution liquid chromatography mass spectrometry (LCMS), using UV and
ELS detection methods run in positive and negative ion mode, of solutions in
microtiter96well plates.
The types of documentation QC review include:
Certificate of Analysis (COA) Chemical Validation-To establish chemica I
identity (chemical name, CAS, MW) from thesupplierlprovided COA.
DSSToxChemical Information Review-Toensureaccurateandconsistent
substance (CAS, name, description) and structure annotation of the generic
chemical (independent of supplier, lot, batch) as part of the DSSTox chemical
registration process.
The results of the analytical QC review include:
An analytical determination and report supplied to EPA in electronic format.
All spectra and supporting information, along with summary purity and
identity confidence score.
Top of Page
Chemical Preparation Process and
Timeline
EPA has developed a step-by-step process to prepare chemica Is for sending to
EPAcontract laboratories and to those requesting the chemical libra ryfor
-------
research purposes. This process ensures that the neat chemical or DMSO
solution stocks a re provided at qua ntities a nd concentrations pri ma rily designed
for high-throughput testing applications, to be run in assays deemed of sufficient
valueto EPA's ToxCast mission.
The process is as follows:
1. Requestors submit an EPA ToxCast Chemical Request Form outlining the
number of chemicals, quantities, and concentrations needed; background
information on the proposed assay testing; pertinent plating details; and the
recipient of the shipment and desired shipment date.
2. The National Center for Computational Toxicology (NCCT) Chemical
Ma nagement reviews the request, prepa res the order, a nd provides it to the
NCCT chemical services contractor.
3. Based on the order, the contractor aliquots, packages, a nd ships chemica I
samples in various formulations (neat, dried down, or plated solutions), with
sufficient qua ntities of dry ice (to ensure no tha wing), to the designated
la boratories or investigators within or outside the U.S.
It ta kes an average of 2-6 weeks (depending on whether chemical stocks are
available for plating) to process a chemical order request from receipt of a
request to shipment of the samples.
The chemical library is typically requested by contractors hired by EPA to screen
the chemicals, EPA research collaborators with a signed agreement and EPA
staff. External organizations requesting the chemical library must agree to
NCCT's policy of only providing blinded plate maps during assay testing. These
organizations should provide EPA NCCT raw or processed screening results prior
to the unblinding of the plates. However, when collaborators have a legitimate
need for unblinding earlier in testing (e.g., for method development only) or
additional details, such as actual concentration, NCCT will provide either
partially or fully unblinded plate maps.
Top of Page
Links
Chemical Lists
ToxCast Assays
Top of Page
-------
ToxCast Data Generation:
ToxCast Assays
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
About ToxCast
ToxCast uses automated chemical screening technologies, called high-
throughput screening assays, to expose living cells or biologica I macromolecules
such as proteins to chemicals. The cells or isolated proteins are then screened for
changes in biological activity that may suggest potential health hazards.
Cell-based in vitro assays use intact cells to measu re cellu la r cha nges in
response to exposu re with the test substa nces. Exa mple cells include hu ma n
prima ry cells a nd cell lines a nd rat prima ry liver cells. An exa mple is
ACEA_T47D.
Biochemical in vitro assays measure the activity of a biological
macromolecule, either a purified protein or a nucleic acid. An example is
NVS_ADME_hCYP19Al.
ToxCast uses more than 700 high-throughput assays that cover a range of high-
level cell responses and approximately 300 signaling pathways.
Click links below for an overview of ToxCast assays and the platform
sources, descriptions, and protocols to test chemical activity in high-throughput
screening.
Assay Information
Overview of ToxCast Assays
ToxCast Assay Platform Sources. Descriptions, and Protocols
Assay Process Overview
-------
Step 2. Data collection
Step 3. Data QC and
analysis
Step 1. High-throughput technology perform repetitive motions of loading the
test chemicals, positive/negative control chemicals, dilution solvents, reagents,
and cells (if applicable} into plates containing numerous wells (e.g., a 384-well
plate).
Step 2. The well plates are incubated for different time durations depending on
the assay. Detection technologies tra nslate the biologica I or physicaI process
i nto a signa I (e.g., fluorescence). The signa Is a re measured a nd compi led
into data bases.
Step 3. Results are quality control checked, normalized, analyzed,
and formatted for release. For more information visit Results Processing
and Analysis.
-------
ToxCast Data Generation:
Overview of ToxCast
Assays
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
About ToxCast
There a re more tha n 15 commercia I or federa I government platform sources
used for ToxCast assays. The ToxCast assays range from simple, single-readout
to complex, multiplexed and multi-parametric technologies.
Each assay ca n generate one or more measured entities or technologica I ta rgets
(ca lied assay components) a nd one or more assay end points (e.g., readouts such
as the concentration at 50% of maximum activity or the AC50). With few
exceptions, the complete ToxCast Phase I and II chemical sets have been tested
across 700+ assay endpoints of data.
The activity of a chemica I in a specific assay does not necessa rily mea n that it
will cause toxicity or an adverse health outcome. There are many factors that
determine whether a chemical will cause a specific adverse health outcome.
Careful review is required to determine the use of the data in a particular
decision context.
The charts a nd ta bles on this page summa rize characteristics of the ToxCast
assays by the following categories:
Assay bv Assay Function Type
Assays by Organism Cell Type
Assay Design Type
Assay by Biological Activity Type
Cell Based Assays bv Cell Line
-------
Assays by Organism Cell Type
NA, 3 bovine,
zebrafish, 2 10
sheep, 2
chimpanzee, 1
Chinese hamster, 1
rabbit, 2
\i
guinea pig, 10
<3 j
¦ bovine
¦ chimpanzee
Chinese hamster
guinea pig
¦ human
¦ mouse
¦pig
¦ rabbit
¦ rat
¦ sheep
¦ zebrafish
Assays by Assay Design Type
growth reporter, 1
viability reporter, 2
morphology reporter. 6
background reportei
membrane potential
reporter, 1
¦ binding reporter
¦ enzyme reporter
inducible reporter
background reporter
¦ morphology reporter
¦ viability reporter
¦ growth reporter
¦ membrane potential reporter
-------
Assays by Biological Activity Type
developmental defect, 2
signaling. 15-
reporter gene, 2<
¦ binding
¦ enzymatic activity
o reporter gene
background control
¦ signaling
¦ developmental defect
¦ viability
Cell Based Assays by Cell Line
III
y yy #///////////
yy y yy y y
J- yy '
-------
ToxCast Data Generation:
Assay Platform Sources,
Descriptions, and
Protocols
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
About ToxCast
Assay Platform Sources
Download ToxCast Assay Platform Source Description document to learn more
about the platform sources of the ToxCast assays.
Assay Descriptions
Descriptions of the ToxCast assays, their components, and end points are
available via the ToxCast dashboard or a downloadable CSV file. This
information is essential for proper interpretation of screening results and
facilitates higher-level analyses.
In the ToxCast Dashboard, click on Choose a View on the top left, select
Assays. Click on the plus sign symbol in front of the assay name to expand the
entry to show the short description of the assay. To see additiona 1 assay
information, click on the assay to highlight it, and click on the Assay Summary
tab on the right side of the dashboa rd.
Assay Summary Information provides descriptions of each ToxCast assay,
which includes its measured readouts and analyzed results.
OECDTest 211 ToxCast Assay Documentation provides descriptions and
guidelines for ToxCast endocrine-related assays in format outlined by the
OECD Guidance Document 211 for describing non-guideline in vitro test
methods.
EPA's ToxCast Assay Annotation Version 1.0 Data User Guide [Version 2.0 Coming
Soon] documents the annotations to assist ToxCast data users in understanding
how to use the annotation terms within the ToxCast screening data. It explains
-------
annotation concepts and terminologies.
Assay Protocols
TOX21
EPA's standard laboratory protocols for Tox21 assays including descriptions,
protocols (reference, quality control, procedures and performance), and assay
data.
National Institutes of Health Tox21 assay descriptions and protocols are also
available on the National Center for Biotechnology Information's PubChem
Bioassav database.
-------
ToxCast Data Generation:
Processing and Analysis
Resources
ToxCast Pipeline ftcph
How to setup pipeline
Overview of R package
ToxCast Pipeline R Package
ToxCast pipeline article
About ToxCast
ToxCast Data
Once the testing is complete, the la bs send the resu Iting data back to EPA for
processing a nd qua lity assura nce/qua lity check!ng. The diversity of the ToxCast
data received from the numerous vendors led to the development of a flexible
high-throughput screening fHTS) data analysis pipeline capable of efficiently
processing and storing large volumes of data. The prescribed procedure EPA
uses to process this data is outlined below, and the pipline used to process
data is available as an R package ftcpl). While developed prima rily for ToxCast,
EPA has attempted to ma ke the tcpl package genera lly a pplicable to the
community for processing any high-throughput chemical screening data.
1. The data, received in unique formats from each vendor, a re transformed to a
sta nda rd computa ble format a nd loaded into the ToxCast data base
(invitrodb) by vendor-specific R scripts.
2. Once data is loaded into the database, the ToxCast Pipeline (tcpl) uses the
generalized processing functions implemented in the package to process,
norma lize, model, qua lify, flag, inspect, a nd visua lize the data.
3. The data resulting from the processing is available for download and in the
ToxCast Dashboard.
-------
ToxCast Data Generation:
ToxCast Pipeline (tcpl)
Resources
ToxCast Pipeline R Package EXIT
ToxCast Pipeline Article
About ToxCast
The ToxCast pipeline (tcpl) is an R package (R Core Team, 2016) that provides
storage, norma lization, dose-response modeling a nd visua lization solutions for
HTS screening efforts. The tcpl package provides functionality for two screening
paradigms:
1. single-concentration screening, intended to only identify potentially active
compounds, and
2. multiple-concentration screening intended to identify potentially active
compounds and estimate the efficacy and potency through dose-response
modeling
Based on the screening pa radigm, the package processes data through a set of
Levels as outlined in the figure below. The tcpl data base stores the sa me raw
information at Level Ofor each screening paradigm. To address both data
storage a nd reproducibility issues the tcpl package interacts with a MySQL
relationa I data base to store a 11 data a nd processing decisions made by the user.
In addition to storing the data at every level of processing, the accompa nying
data base stores chemica I a nd assay a nnotations to facilitate further a na lyses
and disseminating results.
The fina I processed data, including hit ca lis, ca n be found on the ToxCast
Dashboard.
For more details on downloading and using the tcpl, please click here. Detailed
information about all processing steps is included in the package vignette.
Data Analysis Pipeline Overview
-------
r
Single-concentration screening
intended to identify potentially active
compounds
Raw file processing (vendor-specific)
I i
I
Multiple-concentration screening
intended to identify potentially active
compounds and estimate efficacy and
potency through dose-response modeling
Assay endpoint-specific normalization
I
Sample processing and hit-calling
Define replicate and concentration indices
I
Assay component-specific value corrections
i
Assay endpoint-specific normalization
I
Model fitting
i
Model selection and hit-calling
i
Flagging (detect potential false + and false -)
Dose-Response Modeling
The dose-response modeling procedure was designed to handle the outliers and
cytotoxicity-related loss of signal common to HTS efforts. The modeling
methodology evaluates each data series with three models: (i) a constant model
at zero (orange in figure), (ii) a constrained three-parameter Hill model (blue) and
(iii) a constrained five-parameter gain-loss model (red).
tp
1 +I0'9a~*,9,v
n-tpf. 1 -Y. 1
ll +10(9a-*>3"'Al+10(xHi,)'w
Hi = 0
To facilitate simple cross experiment comparisons and reduce the parameter
space, dose-response modeling is constrained to a zero-centered, positive
response paradigm. Therefore, negative response data requires an inverse
transformation during the normalization process. To obtain robust results
without removing any data, we define the log-likelihood using a Student's t
distribution with 4 degrees of freedom (La nge et a L, 1989). Utilizi ng the Student's
t distribution and the gain-loss model provide robust solutions to the problems
of outliers a nd the cytotoxicity related signa I loss that commonly occu rs at high
concentrations. After fitting each model, the model with the lowest Akaike
information criterion value is selected as the'winning' model.
Hit-Calling and Point-of-Departure Estimates
A dose-response series must meet three criteria to have an active hit call: (i) the
Hill or Gain-Loss model must win and (ii) the modeled curve fit top must exceed
the efficacy cutoff, a nd (iii) for at least one concentration the median response
value must also exceed the efficacy cutoff. The efficacy cutoff is a user-defined
value selected from a n expa ndable list of methods at Level 5.
In addition to the sta ndard AC50 (activity concentration at 50% of maxima I
activity) provided for the Hill and Gain-Loss models, the tcpl package provides
-------
three point of departure (POD) estimates for the winning model. The AC10, or
activity concentration at 10%, is derived solely from the model parameters.
Conversely, the ACB (activity concentration at baseline) and ACC (activity
concentration at cutoff) a re based on levels of noise a nd significa nee,
respectively.
The package estimates the noise of an assay by calculating the median absolute
deviation over a II response va lues given by the first two concentrations (bmad).
The baseline region is then defined as 0±3bmad, and the ACB is the concentration
at which the model first reaches 3bmad. Similarly, the ACC is the concentration
at which the model first reaches the user-defined cutoff value for a data-series to
be considered active. The four POD estimates are illustrated in the figure.
No POD estimates a re ca Iculated when the consta nt model is the selected model
winner, because the POD estimates do not apply.
Links
Publicly Available Data
-------
Exploring ToxCast Data:
Overview of tcpl R Package
Resources
Toxicitv Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
ToxCast Citation
About ToxCast
TheToxCast pipeline (tcpl) isan R package that serves as an interface for dose-
response modeling and visualization for high-throughput screening efforts
based on the data generated by ToxCast. For advanced users who would like to
use the tcpl package to process their own data or review the ToxCast data
processing, the R package is availa ble for download exit . Users will need
experience with R to use these files. The R package includes:
A reference manual.
A vignette describing the package in detail.
Exercises to help users work through and understand the vignette.
A list of package functions.
The user ca n choose to use the tcpl package (v2.0) with a MySQL data base
or with comma-delimited (*.csv) files that do not require MySQL. Afull
EXIT
description ofthetcpl configuration options is available in the R manual fortcpl
v2.0.
See How to Set u p R a nd MvSOL if needed before using.
The package includes the following sections to describe the functionality of the
pipeline.
1. Introduction
2. Register and Upload New Data
3. Data Processing and the tcplRun Function
4. Data Normalization
5. Single-concentration Screening
a. Levell
b. Level2
6. Multiple-concentration Screening
a. Levell
b. Level2
c. Level3
d. Level4
e. Level5
f. Level6
7. Field Explanation/Database Structure
-------
a. Single-concentration tables
b. Multiple-concentration tables
c. Auxiliary annotation tables
8. B Level 0 Pre-processing
9. C BurstZ-ScoreCalculation
Links
How to Set U p R a nd MySQL
-------
Exploring ToxCast Data:
How to Set Up R and
MySQL
Resources
Toxicitv Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
ToxCast Citation
About ToxCast
The following links exit the site exit
It is helpful to have a working knowledge of R and MySQL to use the ToxCast tcpl
R package. For further details as to how to use Rand MySQL, please reference
these instructions and external resources.
Setting up R
1. Download the latest version of the R package from the CRAN according to
your operating system (Linux, Mac, Windows),
a. Review the FAQs and instructions for installing R for the first time
(Windows) (Mac) (Linux).
Setting up MySQL
1. Download the latest version of the MvSOL community server
a. Select the appropriate installer for your operating system
i. Windows: MSI installer
ii. MAC: DMG installer
b. The installer will walkyou through the installation
c. During the installation, be sure to copy the temporary root
password. You will need it later.
d. For Windows, MySQL should automatically be added to your PATH.
e. For MAC, if MySQL was not added to your path automatically then add it
manually
i. Open terminal and type:
echo 'export PATH=/usr/local/mysql/bin:$PATH' >>
~/-bash_profile
2. Open the command line (Windows) or terminal (MAC) to login to the MySQL
server with the command:
-------
mysql -u root -p
3. Enter the temporary root password when prompted for a password.
4. Change the root password
a. Full documentation can befound at MvSOL Website
5. Create the data base, select it as the default database, a nd load the dump file:
mysql> CREATE DATABASE IF NOT EXISTS invitrodb_v2j
mysql> USE invitrodb_v2j
mysql> source prod_external_invitrodb_v2.all.sql
a. Further documentation can be found at MvSOL Website
b. Be patient; this can ta ke a nywhere from 3 to 8 hours.
Links
Overview of R Package
-------
ToxCast Data Generation:
Publicly Available Data
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
ToxCast Citation
About ToxCast
Data Releases
TheToxCast effort has been developed in three phases: Phase I, Phase H
and Phase III (in progress). Starting with the release of Phase I data in 2010, each
phase has added new data for unique chemicals and assay end points to the
library.
There is currently data available for more than 1800 unique chemicals and
700 assay endpoints (Note; not all chemicals a re subjected to testing in all
assays). The graphic belo w presentsa breakdown by phase ofthechemicalset
na mes, release dates, number of unique chemica Is assessed a nd number of
assay endpoints.
The current data files a re availa ble for download here.
Phase I
Released: 2010
Chemical set: "phl_vl"
310 chemicals
~700 assay endpoints
Phase II
Released: 2013
Chemical set: "phl_v2"
293 chemicals
~200 assay endpoints
Chemical set: "ph2"
768 chemicals
~900 assay endpoints
Chemical set: "elk"
799 chemicals
~50 assay endpoints
Phase III
Initiated: 2014
r
Chemical set: "ph3"
Assay testing in process
2,678 chemicals
Previous ToxCast Data Releases
Data files from previously published ToxCast data releases a re availa ble
here. These files a re NOT recommended for use in new analyses, but are
provided for users who may need access to previous data files for ongoing
analyses.
Additional Information on Data
Releases
-------
Richard. A. M.. Judson. R. S.. Houck. K.A..Grulke.C. M..Volarath. P..
Thillainadarajah.I &Knudsen.T. B. f2016).ToxCastchemica[landscape:
Paving the road to 21st century toxicology. Chemical Research in Toxicology.
29(8). 1225-1251.
Links
ToxCast Data
Downloadable Data
-------
Exploring ToxCast Data
Resources
Toxicitv Forecaster (ToxCast)
Fact Sheet
DownloadableToxCast Data
ToxCast Dashboard
ToxCast Publications
ToxCast Citation
About ToxCast
Example Use Cases
ToxCast data, once it has been generated by labs and processed by EPA, can be
downloaded from our website a nd is a Iso available in the ToxCast Dashboa rd.
Download ToxCast Data
For more advanced users, the data files themselves are available for download
and use, as well as the data processing pipeline used by EPA (tcpl R package).
Downloadable Data
ToxCast Dashboard
TheToxCast Dashboard was developed by EPA to help users examine high
throughput assay data to inform chemical safety decisions. All of the publicly
released ToxCast data a re accessible via the dashboard in an intuitive and easy-
to-use format. EPA expects that most users will be a ble to find and explore data
of interest using this format. The Dashboard includes Help annotation and
instructions for use.
ToxCast Dashboard
Note: For the best user experience, we recommend using Google Chrome or
Firefox
f=TZ == 1
r ~ -
-------
Acessing ToxCast Data and Options
for Exploring Data
Click here for examples of how ToxCast data can be accessed and explored
by users (through the Dashboard orthetcpl) or how ToxCast has been used
to inform decision making.
-------
Exploring ToxCast Data:
Accessing ToxCast Data
and Scenarios for
Exploring Data
ToxCast data can be accessed, explored and used in different ways and for
different purposes. This page highlights examples of ways to access and explore
the data.
Resources
Toxicitv Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
ToxCast Citation
About ToxCast
On This Page
ToxCastDashboard
R-PackapeTCPL Scenarios
° Explore the dashboard
° Export data
° Explore bioloeical activitv
° Register new data
° Explore chemical activity
° Process mv own data
Endocrine Disruptor Screening Program
Links
ToxCast Dashboard Exploration
Examples
The ToxCast Dashboard provides three useful tutoria Is to demonstrate possible
use cases.
1 would like to explore the dashboard in general.
This tutorial is located under the Chemical Summary tab of the Dashboard and
provides introductory information showing users how to select chemicals in the
Dashboard. Users can follow a similar process to select assays in the Dashboard.
1.
2.
There are four tabs in the top right half of the ToxCast Dashboard home
page, Chemical Summary, Assay Summary, Bioactivity,and Help.
U nder the Chemical Summary tab, scroll down a nd click on the "Sta rt
-------
Tutorial-ChemicalTab" button.
3. Apoplupwindow willappear with instructions. Follow each instruction
before clicking the "Next" button.
4. There is no tutorial under the Assays tab, but users can follow the process
outlined in Chemical tutorial to explore on their own. Be sure to change the
"Choose a view" option in the top left a nd click over to the Assay Summary
ta b to see what type of information is available. For detailed information on
these assays, please see the Assays page.
Top of Page
I would like to explore biological activity.
This tutorial can be found under the Bioactivity tab of the Dashboard. This
tutorial assumes that users are familiar with selecting assays and chemicals in
the Dashboard,
1. Under the Bioactivity tab, click on the "Start Tutorial - Bioactivity Tab"
button.
&ERA iCSS ToxCast Dashboard
l1
mum r
| ^ |
SamptM On*)
CA&liN (luatfat Dm liny 1 Ifi Mn*n MKil
2, Apoplupwindow willappear with instructions. Follow each instruction
before clicking the "Next" button.
-------
3. Users ca n sta rt by selecting either a n assay or a chemica I, depending on their
interest.
4. Users can also download data into a .csvfile.
Top of Page
I would like to explore chemical activity.
Example - Activity in flame retardants.
This tutorial is located under the Help tab of the Dashboard and explores the
data relevant to BisphenolA and all Estrogen Receptor (ER) assays,
1. Under the Help tab, expand the "How to Use the iCSS Dashboard" window.
2. Clickonthe"UseCaseTutorial" button. Maximize your browser window.
SB* iCSS ToxCast Dashboard
3. A poplup window will appear with instructions. Follow each instruction
before clicking the "Next" button. Please note that when you click on the
magnifying glass in the assay window, a new Assay Explorer window will pop
up and may blockyour instructions; if this happens, simply movethe new
window slightly to the right and down a nd press the "Next" button on the
instruction poplup.
-------
P1«M« Click IN
»fi Bit n;
-------
I would like to process my own data using the tcpl
pipeline.
This example assumes that users are familiar with R and the tcpl R Package and
have registered their data (see previous example for details).
This example, available in the tcpl Vignette exit ,explains howthedata are
processed through the levels of the pipeline for single-concentration and
multiple-concentration screening.
AH processing in the tcpl package occurs at the assay component or assay
end point level. There is no capability within either screening paradigm to do any
processing which combines data from multiple assay components or assay
end points. Any combining of data must occur before or after the pipeline
processing.
The processing is sequential, and every level of processing requires successful
processing at the antecedent level. The processing requirements vary by
screening paradigm and level. The Vignette includes the details, but in general,
many of the processing steps require specific methods to accommodate
different experimenta 1 designs or data processing a pproaches.
Seethe tcpl Vignette exit for further details.
Top of Page
ToxCast Data Relevant to EPA's
Endocrine Disruptor Screening
Program
Based on scientific adva nces, EPA intends to implement the use of high
throughput screening assays and computational models to evaluate, and to a
significa nt extent, screen chemica Is.
Endocrine Disruptor Screening Program. The in vitro high throughput and
computational model alternatives provide an accurate quantitative measure
of specific endocrine receptor binding bioactivity and mechanisms that can
serve as a Iternatives to the current Tier 1 estrogen receptor (ER) binding,
ER transactivation (ERTA) and uterotrophic assays.
Prioritizing Chemicals for Potential Endocrine Disruption. EPA resea rchers
have been developing new rapid methods that can quickly screen thousands
of chemicals.
High-throughput Screening Data for Estrogen Receptor Model. Estrogen
receptor model data from the manuscrpit titled Integrated Model of Chemical
Perturbations of Biological Pathways Using 18 In Vitro High-Throughput
Screening Assays for the Estrogen Receptor (Judson et a I.) published in
Toxicological Sciences.
Links
ToxCast Dashboard
-------
Downloadable Data
Top of Page
-------
Exploring ToxCast Data:
Downloadable Data
Resources
Toxicitv Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
ToxCast Citation
About ToxCast
The results after processing through the Pipeline are available on the ToxCast
Dashboa rd. and for most users EPA recommends accessing the data there.
ToxCast Chemicals
ToxCast Assays
ToxCast Data and Information
T oxCast & T ox21 Su m ma rv Fi les. Data for a single chemical endpoint pairfor
thousa nds of chemica Is a nd assay endpoints for 20 va ria bles such as the
activity or hit call, activity concentrations, whether the chemica I was tested
in a specific assay, etc.
ToxCast&Tox21 Data Spreadsheet. A spreadsheet of EPA's analysis of the
chemica Is screened through ToxCast a nd the Tox21 collaboration which
includes EPA's activity ca lis from the screening of over 1,800 chemica Is.
ToxCast & Tox21 Concentration Response Plots. Concentration response
plots for a 11 of the ToxCast a nd Tox21 assays.
Collaborative Estrogen Receptor Activity Prediction Project Data. Data and
supplemental files from CERAPP (A large-scale modeling project) which
demonstrated the efficacy of using predictive computationa I models trained
on high-throughput screeni ng data to eva luate thousa nds of chemica Is
for estrogen-related activity. CERAPP combined multiple models developed
in collaboration with 17 groups in the United States and Europe to predict ER
activity of a common set of 32,464 chemica I structures. Quantitative
structure-activity relationship models a nd docking a pproaches were
employed, to build a tota I of 40 categorica I a nd 8 continuous models for
binding, agonist, and antagonist ER activity.
High-throughput Screening Data for Estrogen Receptor Model. Estrogen
receptor model data from the manuscript titled Integrated Model of Chemical
Perturbations of Biological Pathways Using 18 In Vitro High-throughput
Screening Assays for the Estrogen Receptor f Judson et a I) published in
Toxicological Science.
High-throughput Screening Data for Androgen Receptor Model. Androgen
receptor activity model data.
Previous ToxCast Data Releases. Data files from previously published ToxCast
data releases. We DO NOT recommend using this data for new analyses, but
-------
are providing these files in case users need them for ongoing analyses.
Animal Toxicity Studies: Effects and Endpoints. The Toxicity Reference
Database (ToxRefDB) provides results form across the thousands of animal
toxicity studies including No Effect Level (NEL)/Lowest Effect Level (LEL)
and/or No Observed Adverse Effect Level (NOAEL)/Lowest Observed Adverse
Effect Level (LOAEL).
ToxCast Results and Processing
Descriptions of how data are processed through can be found on the Results
P rocessi ng a nd Ana lysis.
Links
Overview of R Package
How to Set U p R a nd MvSOL
-------
Exploring ToxCast Data:
Citing ToxCast Data
Resources
Toxicity Forecaster (ToxCast)
Fact Sheet
ToxCast Publications
About ToxCast
The following citation format can be used when citing ToxCast data:
General Citation Suggestion
U.S. EPA. Data download year. Data set namefrom database version. Retrieved
from https://www.epa.gov/chemical-research/toxicity-forecaster-toxcasttm-
data on date retrieved. Data release date.
Example Citation
U.S. EPA. 2015. ToxCast&Tox21 Summary Files from invitrodb_v2. Retrieved
from https://www.epa.gov/chemical-research/toxicitv-forecaster-toxcasttm-
data on October 28,2015. Data released October 2015.
------- |