Topics

We provide two software tools for the analysis conducted as a part of WP5 of the NanoInformaTIX project. We focus on two application areas:

Inference for dose-response NanoInformatics data (the NMTox R package).
Modelling dose response NanoInformatics data and estimation of toxic dose level.

For both application the analysis is implemented using the R software.

Analysis of dose-response NanoInformatics data

Testing for monotone trend of NanoInformatics data

We present a case study and the theoretical background of the first part of the analysis which consists of exploratory stage and testing for monotone trend of the Nanoinformatics data. Isotonic regression is used to estimate the response at each dose level and the likelihood ratio test is used to test for monotonicity (with multiplicity adjustment).

Testing for monotone trends

The NMTox R package

The NMTox R package can be used to perform data exploration and identify significant trends in the dose-response relationship of nanomaterial toxicity. The NMTox-package consists of functions that can be used to generate scatter plots and frequency tables, to subset or split the data, to create plots with isotonic means and to perform statistical inference (including the multiplicity adjustment).

The NMTox R package

Modelling dose-response NanoInformatics data

In the second stage of the analysis consists of the estimation of the dose which is toxic to cells (eg. Cell death) in in vitro cell systems, eg. an EC10 value (a dose at which there is an effect on 10% of the cells exposed to that dose). In order to identify those type of doses, a dose-response curve needs to be fitted to the data and based on the curve, an EC10 value can be determined. Several nonlinear models are fitted, such as Log-logistic, Weibull and Gompertz on the Nanomaterials datasets and a model averaging method is used to estimate the ED10 (EDp).

Dose-response modeling

Dose-response modelling of Nanomaterial Toxicity in R

The drc R package () is used for basic modelling and Model averaging is used for parameter estimation. A working case study and the code in R are provided in this page.

Dose-response models for NM toxicity in R

Testing for monotone trends: A Short course

This chapter is focused on inference and covers the following topics:

Bootstrap tests: one sample problem and two samples problem.
Bootstrap tests for correlation and ratios.

Permutation tests: two samples problem.

Short course