Correction: Anastopoulos et al. Multi-Drug Featurization and Deep Learning Improve Patient-Specific Predictions of Adverse Events. Int. J. Environ. Res. Public Health 2021, 18, 2600.

In the original publication [...].

A machine learning approach for prioritizing groundwater testing for per-and polyfluoroalkyl substances (PFAS).

Regulatory agencies are beginning to recognize and regulate per-and polyfluoroalkyl substances (PFAS) as concerning environmental contaminants. In groundwater management, testing and mitigation strategies are desirable, but can be time and cost-intensive processes. As a result, only a fraction of all groundwater wells has been tested for PFAS levels, resulting in potentially extended drinking water exposure to PFAS in the meantime. In this study, we build a series of machine learning models (including linear and random forest regressors) to predict PFAS based on a groundwater dataset from California. These models are used to compare the relative predictive ability of co-contaminant fingerprints, hydrological properties, soil parameters, proximity of airports/military bases, and geospatial data. Additionally, a random forest machine learning model that combines all data types can quantitatively predict the maximum PFAS compound concentration in a well with a Spearman correlation of 0.64 and can discern wells containing concerningly high concentrations of PFAS with an accuracy of 91 % (AUC of 0.90). This approach may have widespread utility for other hazardous anthropogenic compounds in groundwater. Future investigations should evaluate the practicability of using machine learning to prospectively prioritize contaminant testing in groundwater wells.

Multi-Drug Featurization and Deep Learning Improve Patient-Specific Predictions of Adverse Events.

While the clinical approval process is able to filter out medications whose utility does not offset their adverse drug reaction profile in humans, it is not well suited to characterizing lower frequency issues and idiosyncratic multi-drug interactions that can happen in real world diverse patient populations. With a growing abundance of real-world evidence databases containing hundreds of thousands of patient records, it is now feasible to build machine learning models that incorporate individual patient information to provide personalized adverse event predictions. In this study, we build models that integrate patient specific demographic, clinical, and genetic features (when available) with drug structure to predict adverse drug reactions. We develop an extensible graph convolutional approach to be able to integrate molecular effects from the variable number of medications a typical patient may be taking. Our model outperforms standard machine learning methods at the tasks of predicting hospitalization and death in the UK Biobank dataset yielding an R2 of 0.37 and an AUC of 0.90, respectively. We believe our model has potential for evaluating new therapeutic compounds for individualized toxicities in real world diverse populations. It can also be used to prioritize medications when there are multiple options being considered for treatment.

Parallel Single-Cell RNA-Seq and Genetic Recording Reveals Lineage Decisions in Developing Embryoid Bodies.

Early developmental specification can be modeled by differentiating embryonic stem cells (ESCs) to embryoid bodies (EBs), a heterogeneous mixture of three germ layers. Here, we combine single-cell transcriptomics and genetic recording to characterize EB differentiation. We map transcriptional states along a time course and model cell fate trajectories and branchpoints as cells progress to distinct germ layers. To validate this inferential model, we propose an innovative inducible genetic recording technique that leverages recombination to generate cell-specific, timestamp barcodes in a narrow temporal window. We validate trajectory architecture and key branchpoints, including early specification of a primordial germ cell (PGC)-like lineage from preimplantation epiblast-like cells. We further identify a temporally defined role of DNA methylation in this PGC-epiblast decision. Our study provides a high-resolution lineage map for an organoid model of embryogenesis, insights into epigenetic determinants of fate specification, and a strategy for lineage mapping of rapid differentiation processes.

Evolthon: A community endeavor to evolve lab evolution.

In experimental evolution, scientists evolve organisms in the lab, typically by challenging them to new environmental conditions. How best to evolve a desired trait? Should the challenge be applied abruptly, gradually, periodically, sporadically? Should one apply chemical mutagenesis, and do strains with high innate mutation rate evolve faster? What are ideal population sizes of evolving populations? There are endless strategies, beyond those that can be exposed by individual labs. We therefore arranged a community challenge, Evolthon, in which students and scientists from different labs were asked to evolve Escherichia coli or Saccharomyces cerevisiae for an abiotic stress-low temperature. About 30 participants from around the world explored diverse environmental and genetic regimes of evolution. After a period of evolution in each lab, all strains of each species were competed with one another. In yeast, the most successful strategies were those that used mating, underscoring the importance of sex in evolution. In bacteria, the fittest strain used a strategy based on exploration of different mutation rates. Different strategies displayed variable levels of performance and stability across additional challenges and conditions. This study therefore uncovers principles of effective experimental evolutionary regimens and might prove useful also for biotechnological developments of new strains and for understanding natural strategies in evolutionary arms races between species. Evolthon constitutes a model for community-based scientific exploration that encourages creativity and cooperation.

Perturb-Seq: Dissecting Molecular Circuits with Scalable Single-Cell RNA Profiling of Pooled Genetic Screens.

Genetic screens help infer gene function in mammalian cells, but it has remained difficult to assay complex phenotypes-such as transcriptional profiles-at scale. Here, we develop Perturb-seq, combining single-cell RNA sequencing (RNA-seq) and clustered regularly interspaced short palindromic repeats (CRISPR)-based perturbations to perform many such assays in a pool. We demonstrate Perturb-seq by analyzing 200,000 cells in immune cells and cell lines, focusing on transcription factors regulating the response of dendritic cells to lipopolysaccharide (LPS). Perturb-seq accurately identifies individual gene targets, gene signatures, and cell states affected by individual perturbations and their genetic interactions. We posit new functions for regulators of differentiation, the anti-viral response, and mitochondrial function during immune activation. By decomposing many high content measurements into the effects of perturbations, their interactions, and diverse cell metadata, Perturb-seq dramatically increases the scope of pooled genomic assays.

A Multiplexed Single-Cell CRISPR Screening Platform Enables Systematic Dissection of the Unfolded Protein Response.

Functional genomics efforts face tradeoffs between number of perturbations examined and complexity of phenotypes measured. We bridge this gap with Perturb-seq, which combines droplet-based single-cell RNA-seq with a strategy for barcoding CRISPR-mediated perturbations, allowing many perturbations to be profiled in pooled format. We applied Perturb-seq to dissect the mammalian unfolded protein response (UPR) using single and combinatorial CRISPR perturbations. Two genome-scale CRISPR interference (CRISPRi) screens identified genes whose repression perturbs ER homeostasis. Subjecting ∼100 hits to Perturb-seq enabled high-precision functional clustering of genes. Single-cell analyses decoupled the three UPR branches, revealed bifurcated UPR branch activation among cells subject to the same perturbation, and uncovered differential activation of the branches across hits, including an isolated feedback loop between the translocon and IRE1α. These studies provide insight into how the three sensors of ER homeostasis monitor distinct types of stress and highlight the ability of Perturb-seq to dissect complex cellular responses.

Guided Homing of Cells in Multi-Photon Microfabricated Bioscaffolds.

Tissues contain exquisite vascular microstructures, and patterns of chemical cues for directing cell migration, homing, and differentiation for organ development and function. 3D microfabrication by multi-photon photolithography is a flexible, high-resolution tool for generating 3D bioscaffolds. However, the combined fabrication of scaffold microstructure simultaneously with patterning of cues to create both geometrically and chemically defined microenvironments remains to be demonstrated. This study presents a high-speed method for micron-resolution fabrication of scaffold microstructure and patterning of protein cues simultaneously using native scaffold materials. By the simultaneous microfabrication of arbitrary microvasculature geometries, and patterning selected regions of the microvasculature with the homing ligand P-selectin, this study demonstrates adhesion, rolling, and selective homing of cells in defined 3D regions. This novel ability to generate high-resolution geometries replete with patterned cues at high speed enables the construction of biomimetic microenvironments for complex 3D assays of cellular behavior.

The dynamics of interacting folds under biaxial compressive stresses.

The gradual in-plane compression of a solid film bonded to a soft substrate can lead to surface wrinkling and even to the formation of a network of folds for sufficiently high strain. An understanding of how these folds initiate, propagate, and interact with each other is still lacking. In a previous study, we developed an experimental system to observe the wrinkle-to-fold transition of layered elastic materials under biaxial compressive stresses. Here we focus on the dynamic interaction of a pair of propagating folds under biaxial compression. We find experimentally that their behavior is mediated through their tips and depends on the separation of the tips and their angle of interception. When the angle is lower than 45°, the two folds either form a unique fold by the coalescence of their tips when close enough, or bend their trajectories to intersect each other and form a lenticular region in analogy with cracks. When the angle is higher then 45°, the folds simply intersect and form a T-like junction. We rationalize this behavior by conducting numerical simulations to visualize the stress field around the two tips and find that the initial geometric position of the tips primarily determines the final state of the folds.

Single-cell RNA-seq reveals changes in cell cycle and differentiation programs upon aging of hematopoietic stem cells.

Both intrinsic cell state changes and variations in the composition of stem cell populations have been implicated as contributors to aging. We used single-cell RNA-seq to dissect variability in hematopoietic stem cell (HSC) and hematopoietic progenitor cell populations from young and old mice from two strains. We found that cell cycle dominates the variability within each population and that there is a lower frequency of cells in the G1 phase among old compared with young long-term HSCs, suggesting that they traverse through G1 faster. Moreover, transcriptional changes in HSCs during aging are inversely related to those upon HSC differentiation, such that old short-term (ST) HSCs resemble young long-term (LT-HSCs), suggesting that they exist in a less differentiated state. Our results indicate both compositional changes and intrinsic, population-wide changes with age and are consistent with a model where a relationship between cell cycle progression and self-renewal versus differentiation of HSCs is affected by aging and may contribute to the functional decline of old HSCs.

A Genome-wide CRISPR Screen in Primary Immune Cells to Dissect Regulatory Networks.

Finding the components of cellular circuits and determining their functions systematically remains a major challenge in mammalian cells. Here, we introduced genome-wide pooled CRISPR-Cas9 libraries into dendritic cells (DCs) to identify genes that control the induction of tumor necrosis factor (Tnf) by bacterial lipopolysaccharide (LPS), a key process in the host response to pathogens, mediated by the Tlr4 pathway. We found many of the known regulators of Tlr4 signaling, as well as dozens of previously unknown candidates that we validated. By measuring protein markers and mRNA profiles in DCs that are deficient in known or candidate genes, we classified the genes into three functional modules with distinct effects on the canonical responses to LPS and highlighted functions for the PAF complex and oligosaccharyltransferase (OST) complex. Our findings uncover new facets of innate immune circuits in primary cells and provide a genetic approach for dissection of mammalian cell circuits.