Design and Analysis Strategies for Pooling in High Throughput Screening: Application to the Search for a New Anti-Microbial

TL;DR

This work studies several recently proposed pooling construction methods, as well as a variety of pooled high-throughput screening analysis methods, in order to provide guidance to practitioners on which methods to use.

Abstract

A major public health issue is the growing resistance of bacteria to antibiotics. An important part of the needed response is the discovery and development of new antimicrobial strategies. These require the screening of potential new drugs, typically accomplished using high-throughput screening (HTS). Traditionally, HTS is performed by examining one compound per well, but a more efficient strategy pools multiple compounds per well. In this work, we study several recently proposed pooling construction methods, as well as a variety of pooled high-throughput screening analysis methods, in order to provide guidance to practitioners on which methods to use. This is done in the context of an application of the methods to the search for new drugs to combat bacterial infection. We discuss both an extensive pilot study as well as a small screening campaign, and highlight both the successes and challenges of the pooling approach.

Figures (10)

• Figure 1: Comparison of three pool construction methods by simulation. In all cases, $\sigma=1$, there is $\lceil 0.01k \rceil$ active factors, and hits are detected using the Gauss-Lasso method of Section \ref{['sec:methods']} with threshold $\sigma/8$.
• Figure 2: Comparison of a subset of the analysis methods in Table \ref{['tab:analysis_methods']}, for CRowS designs only. The methods are: the Elastic Net; Gauss-Lasso with $\tau=\sigma/4$ ($\sigma$ known); Gauss-Lasso with $\tau=0.5 \times \text{max}(|\hat{\beta}_{\lambda=0}|)$ ($\sigma$ unknown); Non-Negative Gauss-Lasso with $\tau=\sigma/4$ ($\sigma$ known); Non-Negative Gauss-Lasso with $\tau=0.5 \times \text{max}(|\hat{\beta}_{\lambda=0}|)$ ($\sigma$ unknown); and the $\lambda$-specific Gauss-Lasso with $\tau_{\lambda}=\text{max}(\hat{\beta}_{\lambda})$.
• Figure 3: Comparison of FPR and TPR in CRowS designs for differing secondary criterion, using the $\lambda$-specific Gauss-Lasso with $\tau_{\lambda}=\text{max}(\hat{\beta}_{\lambda})$ primary thresholding criterion.
• Figure 4: Lasso profile plots for $(n=320, k=640, c=8)$ proof-of-concept CRowS design. The plot annotates the top 10 compounds in terms of magnitude at the smallest $\lambda$.
• Figure 5: Plots of the data for $(n=320, k=640, c=8)$ proof-of-concept CRowS design. The control wells consist of Carbenicillin as a positive control and DMSO (no compounds) as a negative control.