Machine learning and image-based profiling in drug discovery

Abstract

The increase in imaging throughput, new analytical frameworks and high-performance computational resources open new avenues for data-rich phenotypic profiling of small molecules in drug discovery. Image-based profiling assays assessing single-cell phenotypes have been used to explore mechanisms of action, target efficacy and toxicity of small molecules. Technological advances to generate large data sets together with new machine learning approaches for the analysis of high-dimensional profiling data create opportunities to improve many steps in drug discovery. In this review, we will discuss how recent studies applied machine learning approaches in functional profiling workflows with a focus on chemical genetics. While their utility in image-based screening and profiling is predictably evident, examples of novel insights beyond the status quo based on the applications of machine learning approaches are just beginning to emerge. To enable discoveries, future studies also need to develop methodologies that lower the entry barriers to high-throughput profiling experiments by streamlining image-based profiling assays and providing applications for advanced learning technologies such as easy to deploy deep neural networks.

The drug-induced phenotypic landscape of colorectal cancer organoids

“…High-throughput phenotypic screening of cells with automated microscopy has more recently become an important tool in cell and molecular biology to systematically cluster perturbations according to changes in cell morphology and thereby dissect molecular mechanisms [10][11][12] . For drug discovery, this can be used to decipher the mode-of-action or off-target effects of novel drug candidates [12][13][14]28 . Furthermore, using RNAi or genetic perturbations together with high-throughput imaging, gene function, cellular pathways or genes altered in cancer or other conditions have been analyzed 11 .…”

Searching for Small Molecules with an Atomic Sort

“…

The discovery of biologically active small molecules requires sifting through large amounts of data to identify unique or unusual arrangements of atoms.H ere,w ed evelop, test and evaluate an atom-based sort to identify novel features of secondary metabolites and demonstrate its use to evaluate novelty in marine microbial and sponge extracts.T his study outlines an important ongoing advance towards the translation of autonomous systems to identify,a nd ultimately elucidate, atomic noveltyw ithin ac omplex mixture of small molecules.One of the most critical aspects in the discovery of biologically active small molecules is the elucidation of small molecular motifs with unique three-dimensional displays.T he combination of this process with detailed targetbased mode of action research [1] lies at the foundation of drug lead [2] discovery.W hile automation, [3] miniaturization, [4] digital networking [5] and machine learning-guided high-throughput screening [6] have produced active leads,t he bulk of screening efforts still follow ac entral approach that begins with am olecular ensemble,e ither an extract containing natural products or asmart library of synthetic compounds. [7] Although both synthetic and natural approaches appear different, they typically apply ac ombination of molecular, cellular, or phenotypic screens.W hile effective,s uch approaches are often cluttered by the discovery of redundant structural features and motifs.T his strategy has prevailed, in part, due to our inability to search for structural novelty.Mass spectrometry (MS) methods,a nd associated profiling systems,p rovide an excellent means to characterize molecules,b ut are typically limited to databased compounds [*] Dr.

…”

“…One of the most critical aspects in the discovery of biologically active small molecules is the elucidation of small molecular motifs with unique three-dimensional displays.T he combination of this process with detailed targetbased mode of action research [1] lies at the foundation of drug lead [2] discovery.W hile automation, [3] miniaturization, [4] digital networking [5] and machine learning-guided high-throughput screening [6] have produced active leads,t he bulk of screening efforts still follow ac entral approach that begins with am olecular ensemble,e ither an extract containing natural products or asmart library of synthetic compounds. [7] Although both synthetic and natural approaches appear different, they typically apply ac ombination of molecular, cellular, or phenotypic screens.W hile effective,s uch approaches are often cluttered by the discovery of redundant structural features and motifs.T his strategy has prevailed, in part, due to our inability to search for structural novelty.…”

Improving drug discovery using image-based multiparametric analysis of epigenetic landscape

“…Over the past decade the decreasing cost and remarkable scalability have made high content screening particularly attractive for drug discovery. More recently, novel image analysis coupled with multiparametric analysis and machine learning have significantly impacted our ability to understand and process phenotypic screening outputs (5,6). Despite these advantages, such assays have not been adapted to extract and utilize information for the cellular epigenetic landscape.…”