Supplementary data for "Tia Tate, Grace Patlewicz, Imran Shah, A comparison of machine learning approaches for predicting hepatotoxicity potential using chemical structure and targeted transcriptomic data, Computational Toxicology, Volume 29, 2024, 100301, ISSN 2468-1113, https://doi.org/10.1016/j.comtox.2024.100301.". This dataset is associated with the following publication: Tate, T., G. Patlewicz, and I. Shah. A comparison of machine learning approaches for predicting hepatotoxicity potential using chemical structure and targeted transcriptomic data. Computational Toxicology. Elsevier B.V., Amsterdam, NETHERLANDS, 29: 100301, (2024).

This dataset is a project file generated by BMDExpress 2.2 SW (Sciome, Research Triangle Park, NC). It contains gene expression data for livers of rats exposed to 4 chemicals (crude MCHM, neat MCHM, DMPT, p-toluidine) and kidneys of rats exposed to PPH. The project file includes normalized expression data (GeneChip Rat 230 2.0 Array) using 7 different pre-processing methods (RMA, GCRMA, MAS5.0, MAS5.0_noA calls, PLIER, PLIER16, and PLIER16_noA calls); differentially expressed probe-sets detected by William's method (p<0.05, and minimum fold change of 1.5); probeset-level and pathway-level BMD and BMDL values from transcriptomic dose-response modeling. This dataset is associated with the following publication: Mezencev, R., and S. Auerbach. The sensitivity of transcriptomics BMD modeling to the methods used for microarray data normalization. PLOS ONE. Public Library of Science, San Francisco, CA, USA, 15(5): e0232955, (2020).

Datasets used in RD-023418: Adverse Outcome Pathway-Driven Identification of Rat Hepatocarcinogens in Short-Term Assays. This dataset is associated with the following publication: Rooney, J., T. Hill, C. Qin, F. Sistare, and C. Corton. Adverse outcome pathway-driven identification of rat liver tumorigens in short-term assays. TOXICOLOGY AND APPLIED PHARMACOLOGY. Academic Press Incorporated, Orlando, FL, USA, 356: 99-113, (2018).

Datasets used in RD-023418: Adverse Outcome Pathway-Driven Identification of Rat Hepatocarcinogens in Short-Term Assays. This dataset is associated with the following publication: Rooney, J., T. Hill, C. Qin, F. Sistare, and C. Corton. Adverse outcome pathway-driven identification of rat liver tumorigens in short-term assays. TOXICOLOGY AND APPLIED PHARMACOLOGY. Academic Press Incorporated, Orlando, FL, USA, 356: 99-113, (2018).

Gene expression data from the Fluidigm qRT-PCR arrays was analyzed in R (v3.6.1; R Foundation for Statistical Computing, 2019). Prior to processing through the tcpl package, each qRT-PCR primer set was annotated as an individual assay endpoint (aeid) for analyses. For each plate, well types were designated for test compound wells (t), positive controls (c), (that is phenobarbital) and neutral controls (n, DMSO). Fold-change in the number of amplification cycles needed to pass the background threshold (Ct) for 96 transcripts to (ftp://newftp.epa.gov/COMPTOX/CCTE_Publication_Data/CCED_Publication_Data/Wambaugh/ToxCast_LTEA, file LTEA_Level2_20191119.zip) were normalized to the geometric mean of three housekeeping genes (ACTB, GAPDH, POLR2A) to generate ΔCt values (cval). Prior to calculating the response values (rval), or ΔΔCt, for each transcript (n = 96) per well, the baseline value (bval), the plate-wise median of the neutral control wells, was generated for each plate (the normalization process is described in detail in supplemental file SupFile4-DeltaCTCalculation.docx). The bval was subtracted from the cval to yield the rval or log2 Fold Change per transcript. Gene expression data from the Fluidigm qRT-PCR arrays was analyzed in R (v3.6.1; R Foundation for Statistical Computing, 2019). Prior to processing through the tcpl package, each qRT-PCR primer set was annotated as an individual assay endpoint (aeid) for analyses. For each plate, well types were designated for test compound wells (t), positive controls (c), (that is phenobarbital) and neutral controls (n, DMSO). Fold-change in the number of amplification cycles needed to pass the background threshold (Ct) for 96 transcripts to (ftp://newftp.epa.gov/COMPTOX/CCTE_Publication_Data/CCED_Publication_Data/Wambaugh/ToxCast_LTEA, file LTEA_Level5_20191119.zip) were normalized to the geometric mean of three housekeeping genes (ACTB, GAPDH, POLR2A) to generate ΔCt values (cval). Prior to calculating the response values (rval), or ΔΔCt, for each transcript (n = 96) per well, the baseline value (bval), the plate-wise median of the neutral control wells, was generated for each plate (the normalization process is described in detail in supplemental file SupFile4-DeltaCTCalculation.docx). The bval was subtracted from the cval to yield the rval or log2 Fold Change per transcript. Supplemental File LTEA_Inucyte_Images.zip is comprised of 20,493 images totaling more than 15 gigabytes. Cell morphology images were acquired for each well/plate with an Essen IncuCyte™ FLR automated phase-contrast microscope located inside a tissue culture incubator. Six 96-well culture plates were loaded into the instrument and imaged for an elapsed time (~24 minutes). The IncuCyte™ software was used for image capturing and export of images in JPEG format. This dataset is associated with the following publication: Franzosa, J., J. Bonzo, J. Jack, N.C. Baker, P. Kothiya, R. Witek, P. Hurban, S. Siferd, S. Hester, I. Shah, S. Ferguson, K. Houck, and J. Wambaugh. High-throughput toxicogenomic screening of chemicals in the environment using metabolically competent hepatic cell cultures. npj Systems Biology and Applications. Springer Nature Group, New York, NY, 7: Article 7, (2021).

Gene expression data from the Fluidigm qRT-PCR arrays was analyzed in R (v3.6.1; R Foundation for Statistical Computing, 2019). Prior to processing through the tcpl package, each qRT-PCR primer set was annotated as an individual assay endpoint (aeid) for analyses. For each plate, well types were designated for test compound wells (t), positive controls (c), (that is phenobarbital) and neutral controls (n, DMSO). Fold-change in the number of amplification cycles needed to pass the background threshold (Ct) for 96 transcripts to (ftp://newftp.epa.gov/COMPTOX/CCTE_Publication_Data/CCED_Publication_Data/Wambaugh/ToxCast_LTEA, file LTEA_Level2_20191119.zip) were normalized to the geometric mean of three housekeeping genes (ACTB, GAPDH, POLR2A) to generate ΔCt values (cval). Prior to calculating the response values (rval), or ΔΔCt, for each transcript (n = 96) per well, the baseline value (bval), the plate-wise median of the neutral control wells, was generated for each plate (the normalization process is described in detail in supplemental file SupFile4-DeltaCTCalculation.docx). The bval was subtracted from the cval to yield the rval or log2 Fold Change per transcript. Gene expression data from the Fluidigm qRT-PCR arrays was analyzed in R (v3.6.1; R Foundation for Statistical Computing, 2019). Prior to processing through the tcpl package, each qRT-PCR primer set was annotated as an individual assay endpoint (aeid) for analyses. For each plate, well types were designated for test compound wells (t), positive controls (c), (that is phenobarbital) and neutral controls (n, DMSO). Fold-change in the number of amplification cycles needed to pass the background threshold (Ct) for 96 transcripts to (ftp://newftp.epa.gov/COMPTOX/CCTE_Publication_Data/CCED_Publication_Data/Wambaugh/ToxCast_LTEA, file LTEA_Level5_20191119.zip) were normalized to the geometric mean of three housekeeping genes (ACTB, GAPDH, POLR2A) to generate ΔCt values (cval). Prior to calculating the response values (rval), or ΔΔCt, for each transcript (n = 96) per well, the baseline value (bval), the plate-wise median of the neutral control wells, was generated for each plate (the normalization process is described in detail in supplemental file SupFile4-DeltaCTCalculation.docx). The bval was subtracted from the cval to yield the rval or log2 Fold Change per transcript. Supplemental File LTEA_Inucyte_Images.zip is comprised of 20,493 images totaling more than 15 gigabytes. Cell morphology images were acquired for each well/plate with an Essen IncuCyte™ FLR automated phase-contrast microscope located inside a tissue culture incubator. Six 96-well culture plates were loaded into the instrument and imaged for an elapsed time (~24 minutes). The IncuCyte™ software was used for image capturing and export of images in JPEG format. This dataset is associated with the following publication: Franzosa, J., J. Bonzo, J. Jack, N.C. Baker, P. Kothiya, R. Witek, P. Hurban, S. Siferd, S. Hester, I. Shah, S. Ferguson, K. Houck, and J. Wambaugh. High-throughput toxicogenomic screening of chemicals in the environment using metabolically competent hepatic cell cultures. npj Systems Biology and Applications. Springer Nature Group, New York, NY, 7: Article 7, (2021).

Dataset includes overview text document (accepted version of manuscript) and tables, figures, and supplementary materials. Supplementary tables provide summary data underlying figures, as noted in the text. This dataset is associated with the following publication: Wehmas, L., A. Deangelo, S. Hester, B. Chorley, G. Carswell, G. Olson, M. George, J. Carter, S. Eldridge, A. Fisher, B. Vallanat, and C. Wood. Metabolic Disruption Early in Life is Associated With Latent Carcinogenic Activity of Dichloroacetic Acid in Mice. TOXICOLOGICAL SCIENCES. Society of Toxicology, 159(2): 354-365, (2017).

This dataset provides measured in vitro intrinsic clearance rates for 54 chemicals tested using isolated hepatocytes from humans, rats, and rainbow trout. The test chemicals were selected to provide broad coverage across the industrial and pesticidal chemical space while also prioritizing chemicals of interest to EPA’s Endocrine Disruptor Screening Program (EDSP). A data evaluation framework was developed to identify results suitable for rate reporting. Acceptable results were then used to evaluate the chemical domain of applicability of the applied methods, the influence of starting substrate concentration on measured rates of intrinsic clearance, and differences in metabolic activity among species. These findings provide data for chemicals of specific interest to the EDSP. More importantly, the results provide critical guidance on future use of in vitro biotransformation assays to support high-throughput chemical risks assessments. This dataset is associated with the following publication: Black, S., J. Nichols, K. Fay, S. Matten, and S. Lynn. Evaluation and comparison of in vitro intrinsic hepatic clearance rates measured using cryopreserved hepatocytes from humans, rats, and rainbow trout. TOXICOLOGY. Elsevier Science Ltd, New York, NY, USA, 457: 152819, (2021).

This dataset provides measured in vitro intrinsic clearance rates for 54 chemicals tested using isolated hepatocytes from humans, rats, and rainbow trout. The test chemicals were selected to provide broad coverage across the industrial and pesticidal chemical space while also prioritizing chemicals of interest to EPA’s Endocrine Disruptor Screening Program (EDSP). A data evaluation framework was developed to identify results suitable for rate reporting. Acceptable results were then used to evaluate the chemical domain of applicability of the applied methods, the influence of starting substrate concentration on measured rates of intrinsic clearance, and differences in metabolic activity among species. These findings provide data for chemicals of specific interest to the EDSP. More importantly, the results provide critical guidance on future use of in vitro biotransformation assays to support high-throughput chemical risks assessments. This dataset is associated with the following publication: Black, S., J. Nichols, K. Fay, S. Matten, and S. Lynn. Evaluation and comparison of in vitro intrinsic hepatic clearance rates measured using cryopreserved hepatocytes from humans, rats, and rainbow trout. TOXICOLOGY. Elsevier Science Ltd, New York, NY, USA, 457: 152819, (2021).

This repository contains the necessary data, python source code and jupyter notebooks to reproduce the results from our manuscript, "Estimating Hepatotoxic Doses Using High-content Imaging in Primary Hepatocytes." Using in vitro data to estimate point of departure (POD) values is an important component of new approach method (NAM)-based chemical risk assessments. In this case study we evaluated a NAM for hepatotoxicity based on rat primary hepatocytes, high-content imaging (HCI) and in vitro to in vivo extrapolation (IVIVE). This dataset is associated with the following publication: Shah, I., T. Antonijevic, B. Chambers, J. Harrill, and R. Thomas. Estimating Hepatotoxic Doses Using High-content Imaging in Primary Hepatocytes. TOXICOLOGICAL SCIENCES. Society of Toxicology, RESTON, VA, 183(2): 285-301, (2021).