Quantifying uncertainty of tuberculosis drug susceptibility range from single-microplate test

TL;DR

This paper introduces a method to quantify uncertainty in drug susceptibility tests for tuberculosis using single-microplate experiments.

Contribution

A novel method for combinatorial enhancement and non-parametric statistical processing of single-plate data to report uncertainty ranges.

Findings

A numerical uncertainty range is quantified from single-plate experiments using non-parametric statistics.

The method is demonstrated on ten anti-tuberculosis drugs using the H37Rv strain of Mycobacterium tuberculosis.

The approach avoids overfitting by using a four-parametric Hill’s function.

Abstract

•A numerical uncertainty range quantification from single-plate experiments.•Overcome possible overfitting with the four-parametric Hill’s function.•Processed dataset for first- and second-line antimycobacterial drugs.

A numerical uncertainty range quantification from single-plate experiments.

Overcome possible overfitting with the four-parametric Hill’s function.

Processed dataset for first- and second-line antimycobacterial drugs.

The registration of fluorescence intensity converted from resazurin in the course of the respective microtiter assay, as a response to the activity of antibacterial drugs, provides a basis for the quantitative characterisation of their activity. At the same time, this assay operates with relatively small samples, and the studied biochemical process exhibits variability. In this work, we address the question of reporting the minimal inhibitory concentration arguing in favour of its range rather than a single value. To achieve this goal, we propose a method for the combinatorial enhancement of single microplate-based data, followed by non-parametric statistical processing. The approach is illustrated by the case study of ten first- and second-line anti-tuberculosis drugs acting on the standard laboratory strain H37Rv of Mycobacterium tuberculosis.

Figures