U.S. flag

An official website of the United States government

Dot gov

The .gov means it's official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.


The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Biomarker Description

TGx-DDI Biomarker Description

  • Developed in human TK6 cells
  • The gene set was derived from TK6 cells exposed to a training set of prototypical DNA damage-inducing agents and chemicals with a clean genetic toxicology profile (28 chemicals: 13 DNA damage-inducing, 15 non DNA-damage inducing)
  • Comprised of 64 gene probes
  • Critical to have intact p53 response as many of the genes are regulated by p53
  • Measures early responses to DNA damage-inducing agents and classifies agents as DNA damage-inducing (DDI) or non-DNA damage-inducing (NDDI)
  • For direct-acting chemicals (i.e., not requiring metabolic activation), recommend 4 hours of exposure and immediate sampling in TK6 cells
  • Demonstrated to work in the presence of various types of rat microsomal liver S9, but recommend keeping %S9 as low as possible
  • For chemicals requiring metabolic activation, recommend 4 hours of exposure in presence of S9 following by 3-4 hours of recovery period
  • Dose-optimization protocols described in Li et al., 2015 and Buick et al., 2015
  • Recommended positive control: X-ray, cisplatin, benzo(a)pyrene
  • Developed on the Agilent Whole Genome microarray platform. This platform is expected to work best with datasets generated from Agilent microarrays.

Things to know about using the classifier

  • This classifier supports the following platforms:
    • Agilent Human Genome 8X60K
      • one dye
      • two dyes
      • dye-swapped
    • Affymetrix Human Genome U133 Plus 2.0 array
    • Generic arrays (must use log2 transformed data for a single chemical and dose)
    • Batch data (must use log2 transformed data for multiple chemicals and/or doses)
  • Data must be in the following formats based on the platform selected:
    • Agilent: delimited text file (.txt)
    • Affymetrix: CEL file
    • Generic: delimited text file (.txt)
    • Batch data: Excel (.xlsx)
  • Probe IDs for Generic and Batch data files should be converted to Agilent IDs. This can be achieved using a web based tool such as the Gene ID Conversion Tool from DAVID Bioinformatics Resources
  • Data in the Results table remain available up to 72 hours or until removed

Classification Process

The TGx-DDI Biomarker and the processes underlying data analysis for classification are described in detail by Jackson et al., 2017. Briefly, the probability that test data fit the profile of DDI or NDDI compounds is determined based on similarity of the test chemical expression profile to the reference chemical profiles. The analysis uses the Nearest Shrunken Centroids (NSC) method along with the statistical and bioinformatics tools described by Tibshirani et al., 2002. Heatmaps, hierarchical clustering using Euclidean distances with average linkage, principal component analyses (PCA) using the prcomp function in R software, and the hclust function from R, also are used to aid in the interpretation of the results. Principle components are estimated using only the training data set. If a chemical is deemed positive for any one of three prediction analyses (NSC heatmaps, PCA, or 2DC ), it is classified as DDI otherwise it is classified as NDDI.

principle components function in R
hierarchical clustering function in R
two-dimensional clustering