A machine learning approach to define antimalarial drug action from heterogeneous cell-based screens

Abstract

Drug resistance threatens the effective prevention and treatment of an ever-increasing range of human infections. This highlights an urgent need for new and improved drugs with novel mechanisms of action to avoid cross-resistance. Current cell-based drug screens are, however, restricted to binary live/dead readouts with no provision for mechanism of action prediction. Machine learning methods are increasingly being used to improve information extraction from imaging data. These methods, however, work poorly with heterogeneous cellular phenotypes and generally require time-consuming human-led training. We have developed a semi-supervised machine learning approach, combining human- and machine-labeled training data from mixed human malaria parasite cultures. Designed for high-throughput and high-resolution screening, our semi-supervised approach is robust to natural parasite morphological heterogeneity and correctly orders parasite developmental stages. Our approach also reproducibly detects and clusters drug-induced morphological outliers by mechanism of action, demonstrating the potential power of machine learning for accelerating cell-based drug discovery.

Actomyosin forces and the energetics of red blood cell invasion by the malaria parasite Plasmodium falciparum

“…A strategy was developed to express mutant alleles of Pfmyoa from a second locus in the PfMyoA-cKO parasite line, to enable conditional mutation of any part of PfMyoA. We used the p230p locus (PlasmoDB ID PF3D7_0208900), identified as dispensable throughout the parasite life cycle [ 44 ] and developed for targeted CRISPR/Cas9 integration [ 45 ]. To facilitate gene replacement on top of the PfMyoA-cKO background, the p230p targeting plasmid (pDC2-p230p-hDHFR) was modified by the exchange of hdhfr for bsd , which encodes the blasticidin-S-deaminase resistance gene (since PfMyoA-cKO parasites already express hDHFR) to form the pDC2-p230p-BSD targeting plasmid ( Fig 1A ).…”

“…For modification of the p230p locus in the PfMyoA-cKO background, targeting and repair constructs were modified from [ 45 ]. The targeting construct, pDC2-p230p-hDHFR, (originally adapted from [ 66 ]) carries 3xHA-tagged Cas9 and the p230p -targeting guide RNA.…”

“…The bsd sequence was amplified from pB-CBHALO [ 67 ], with the N-terminal sequence modified to MAK during amplification, to match a consensus sequence [ 68 ]. The p230p-BIP-sfGFP repair plasmid (pkiwi003) [ 45 ] was modified to excise the BIP promoter by SacII/NheI digestion, and ligation of a 2.0 kbp region directly upstream of Pfmyoa (roughly 2/3 of the intergenic region) to form p230p-prMA-sfGFP. The Pfmyoa cds with 3xHA-tag was generated synthetically (GeneART) with altered codons and ligated into p230p-prMA-sfGFP after NheI/PstI digestion to form p230p-prMA-PfMyoA.…”

“…Before transfection, plasmids were purified by ethanol precipitation (with 0.1 vol 3M sodium acetate pH 5.2, 1.5 vol 100% ethanol), washed in 70% ethanol, air dried and resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Oligonucleotides used are listed in S1 Table and S2 Table. For modification of the p230p locus in the PfMyoA-cKO background, targeting and repair constructs were modified from [45]. The targeting construct, pDC2-p230p-hDHFR, (originally adapted from [66]) carries 3xHA-tagged Cas9 and the p230p-targeting guide RNA.…”

Use of Artificial Intelligence and Machine Learning for Discovery of Drugs for Neglected Tropical Diseases

“…The abilities of machine learning methods to improve information extraction from imaging data are finding new applications in mechanism of action studies. Because these methods do not work well with heterogeneous cellular phenotypes and require human training, Ashdown and coworkers reported a combined human- and machine-labelled approach for data from mixed human malaria parasite cultures ( Ashdown et al, 2020 ). Trained on high-throughput and high-resolution screening data, their approach tolerates natural parasite morphological heterogeneity and correctly orders parasite developmental stages.…”