High Throughput Screening Quality Control

The robustness of high-throughput screening (HTS) data is paramount to the identification of high-quality small-molecule candidates for further, more comprehensive investigation. Quality control (QC) is a critical aspect of screening which permits the identification of systematic errors and random artefacts, directly influencing the quality of putative hits.

This script is designed to analyse an absorbance-based screening dataset and output several quality metrics and visualisations which can be used to identify the robustness of the screen and any potential issues.

The example dataset provided is a resazurin-based screen of a Nalm6 B-cell Acute Lymphoblastic Leukaemia (ALL) cell line performed in 384-well format with an FDA-approved compound library of 1971 compounds. The plate data was acquired with a plate reader, and raw results were exported in list format to comma-separated value (CSV) files.

Analyses

This script will allow visualisation and calculation of key HTS quality control metrics including:

• Experiment-wide raw absorbances
• Experiment-wide column effects
• Experiment-wide row effects
• Control well variability
• Per-plate descriptive statistics (mean, standard deviation, median, median absolute deviation) for compounds, positive controls, and negative controls.
• Per-plate regression model of positive and negative controls
• Per-plate signal-to-background ratio
• Per-plate Z' factor
• Per-plate robust Z' factor
• Heatmaps of each plate for visualisation

Statistical metrics are exported as graphical figures and as a dataframe.

Getting Started

• Requirements This contains the required dependencies used throughout the script.
• QC_analysis.ipynb Is the easiest way to start. It shows an example of processing a screening dataset derived from an FDA-approved compound library using resazurin fluorescence as the end output.
• HTS_analysis.py Is the script to analyse the data.
• Raw Data Should contain any raw data files for analysis as list format and saved as a Comma-Separated Values (CSV) file. One datafile corresponds to one plate.
• control_locations.csv Should contain information about the position of positive and negative controls located on every plate loaded (POS or NEG). For example:

Well Row	Well Col	COMP_TYPE
A	23	NEG
A	24	NEG
C	23	POS
C	24	POS

Dependencies

The script relies on the following dependencies (tested version provided in parentheses):

• python (3.8)
• pandas (1.1.4)
• matplotlib (3.3.3)
• seaborn (0.11.0)

Install

Clone this repository: git clone https://github.com/hayden-bell/HTS-QC