Systematic reconstruction of molecular pathway signatures using scalable single-cell perturbation screens
Related Articles
A genome-wide atlas of human cell morphology
A key challenge of the modern genomics era is developing empirical data-driven representations of gene function. Here we present the first unbiased morphology-based genome-wide perturbation atlas in human cells, containing three genome-wide genotype–phenotype maps comprising CRISPR–Cas9-based knockouts of >20,000 genes in >30 million cells. Our optical pooled cell profiling platform (PERISCOPE) combines a destainable high-dimensional phenotyping panel (based on Cell Painting) with optical sequencing of molecular barcodes and a scalable open-source analysis pipeline to facilitate massively parallel screening of pooled perturbation libraries. This perturbation atlas comprises high-dimensional phenotypic profiles of individual cells with sufficient resolution to cluster thousands of human genes, reconstruct known pathways and protein–protein interaction networks, interrogate subcellular processes and identify culture media-specific responses. Using this atlas, we identify the poorly characterized disease-associated TMEM251/LYSET as a Golgi-resident transmembrane protein essential for mannose-6-phosphate-dependent trafficking of lysosomal enzymes. In sum, this perturbation atlas and screening platform represents a rich and accessible resource for connecting genes to cellular functions at scale.
Decoding heterogeneous single-cell perturbation responses
Understanding how cells respond differently to perturbation is crucial in cell biology, but existing methods often fail to accurately quantify and interpret heterogeneous single-cell responses. Here we introduce the perturbation-response score (PS), a method to quantify diverse perturbation responses at a single-cell level. Applied to single-cell perturbation datasets such as Perturb-seq, PS outperforms existing methods in quantifying partial gene perturbations. PS further enables single-cell dosage analysis without needing to titrate perturbations, and identifies ‘buffered’ and ‘sensitive’ response patterns of essential genes, depending on whether their moderate perturbations lead to strong downstream effects. PS reveals differential cellular responses on perturbing key genes in contexts such as T cell stimulation, latent HIV-1 expression and pancreatic differentiation. Notably, we identified a previously unknown role for the coiled-coil domain containing 6 (CCDC6) in regulating liver and pancreatic cell fate decisions. PS provides a powerful method for dose-to-function analysis, offering deeper insights from single-cell perturbation data.
Identifying perturbations that boost T-cell infiltration into tumours via counterfactual learning of their spatial proteomic profiles
Cancer progression can be slowed down or halted via the activation of either endogenous or engineered T cells and their infiltration of the tumour microenvironment. Here we describe a deep-learning model that uses large-scale spatial proteomic profiles of tumours to generate minimal tumour perturbations that boost T-cell infiltration. The model integrates a counterfactual optimization strategy for the generation of the perturbations with the prediction of T-cell infiltration as a self-supervised machine learning problem. We applied the model to 368 samples of metastatic melanoma and colorectal cancer assayed using 40-plex imaging mass cytometry, and discovered cohort-dependent combinatorial perturbations (CXCL9, CXCL10, CCL22 and CCL18 for melanoma, and CXCR4, PD-1, PD-L1 and CYR61 for colorectal cancer) that support T-cell infiltration across patient cohorts, as confirmed via in vitro experiments. Leveraging counterfactual-based predictions of spatial omics data may aid the design of cancer therapeutics.
The transcriptomic architecture of common cancers reflects synthetic lethal interactions
To maintain cell fitness, deleterious genetic alterations are buffered by compensatory changes in additional genes. In cancer, buffering processes could be targeted by synthetic lethality. However, despite the large-scale identification of synthetic lethal effects in preclinical models, evidence that these operate clinically is limited. This impedes the application of synthetic lethal approaches. By integrating molecular profiling data from >9,000 cancers with synthetic lethal screens, we show that transcriptomic buffering of tumor suppressor gene (TSG) loss by hyperexpression of synthetic lethal partners is a common phenomenon, extending to multiple TSGs and histotypes. Transcriptomic buffering is also notable in cancers that phenocopy TSG loss, such as BRCAness cancers, where expression of BRCA1/2 synthetic lethal genes correlates with clinical outcome. Synthetic lethal genes that exhibit transcriptomic buffering also represent more robust synthetic lethal effects. These observations have implications for understanding how tumor cells tolerate TSG loss, in part explain transcriptomic architectures in cancer and provide insight into target selection.
A computational spectrometer for the visible, near, and mid-infrared enabled by a single-spinning film encoder
Computational spectrometers enable low-cost, in-situ, and rapid spectral analysis, with applications in chemistry, biology, and environmental science. Traditional filter-based spectral encoding approaches typically use filter arrays, complicating the manufacturing process and hindering device consistency. Here we propose a computational spectrometer spanning visible to mid-infrared by combining the Single-Spinning Film Encoder (SSFE) with a deep learning-based reconstruction algorithm. Optimization through particle swarm optimization (PSO) allows for low-correlation and high-complexity spectral responses under different polarizations and spinning angles. The spectrometer demonstrates single-peak resolutions of 0.5 nm, 2 nm, 10 nm, and dual-peak resolutions of 3 nm, 6 nm, 20 nm for the visible, near, and mid-infrared wavelength ranges. Experimentally, it shows an average MSE of 1.05 × 10⁻³ for narrowband spectral reconstruction in the visible wavelength range, with average center-wavelength and linewidth errors of 0.61 nm and 0.56 nm. Additionally, it achieves an overall 81.38% precision for the classification of 220 chemical compounds, showcasing its potential for compact, cost-effective spectroscopic solutions.
Responses