scDrug: From single-cell RNA-seq to drug response prediction

Machine learning to dissect perturbations in complex cellular systems

Understanding the responses of biological systems to various perturbations, such as genetic, chemical, or environmental challenges, is essential for reconstructing causal network models. Emerging single-cell technologies have become instrumental in elucidating cell states and phenotypes and they have been used in combination with genetic screening. Recent advances in machine learning and artificial intelligence architectures have stimulated the development of computational tools for modeling perturbations and the response to compounds. This study outlined core principles underpinning perturbation analysis and discussed the methodologies and analytical frameworks used to decode drug and genetic perturbation responses, complex multicellular interactions, and network dynamics. The current tools used for various applications were overviewed. These developments hold great promise for improving drug development and personalized medicine. Foundation models and perturbation cell and tissue atlases offer immense potential for advancing our understanding of cellular behavior and disease mechanisms.

A Multimodal Framework to Uncover Drug-Responsive Subpopulations in Triple-Negative Breast Cancer

Understanding how individual cancer cells adapt to drug treatment is a fundamental challenge limiting precision medicine cancer therapy strategies. While single-cell technologies have advanced our understanding of cellular heterogeneity, efforts to connect the behavior of individual cells to broader tumor drug responses and uncover global trends across diverse systems remain limited. There is a growing availability of single-cell and bulk omics data, but a lack of centralized tools and repositories makes it difficult to study drug response globally, especially at the level of single-cell adaptation. To address this, we present a multimodal framework that integrates bulk and single-cell treated and untreated transcriptomics data to identify drug responsive cell populations in triple-negative breast cancer (TNBC). Our framework leverages population-scale bulk transcriptomics data from TNBC samples to define seven main "identities", each representing unique combinations of biologically relevant genes. These identities are dynamic and trackable, allowing us to map them onto single cells and uncover global patterns of how cell populations respond to drug treatment. Unlike static classifications, this approach captures the evolving nature of cellular states, revealing that a select few identities dominate and drive population-level responses during treatment. Crucially, our ability to decode these trends through the inherent noise of single-cell data provides a clearer picture of how heterogeneous cell populations adapt to therapy. By identifying the dominant identities and their dynamics, we can better predict how entire tumors respond to treatment. This insight is essential for designing precise combination therapies tailored to the unique heterogeneity of patient tumors, addressing the single-cell variations that ultimately determine therapeutic outcomes.

Interpreting single-cell and spatial omics data using deep neural network training dynamics

Single-cell and spatial omics datasets can be organized and interpreted by annotating single cells to distinct types, states, locations or phenotypes. However, cell annotations are inherently ambiguous, as discrete labels with subjective interpretations are assigned to heterogeneous cell populations on the basis of noisy, sparse and high-dimensional data. Here we developed Annotatability, a framework for identifying annotation mismatches and characterizing biological data structure by monitoring the dynamics and difficulty of training a deep neural network over such annotated data. Following this, we developed a signal-aware graph embedding method that enables downstream analysis of biological signals. This embedding captures cellular communities associated with target signals. Using Annotatability, we address key challenges in the interpretation of genomic data, demonstrated over eight single-cell RNA sequencing and spatial omics datasets, including identifying erroneous annotations and intermediate cell states, delineating developmental or disease trajectories, and capturing cellular heterogeneity. These results underscore the broad applicability of annotation-trainability analysis via Annotatability for unraveling cellular diversity and interpreting collective cell behaviors in health and disease.

Single-cell transcriptomes identify patient-tailored therapies for selective co-inhibition of cancer clones

Intratumoral cellular heterogeneity necessitates multi-targeting therapies for improved clinical benefits in advanced malignancies. However, systematic identification of patient-specific treatments that selectively co-inhibit cancerous cell populations poses a combinatorial challenge, since the number of possible drug-dose combinations vastly exceeds what could be tested in patient cells. Here, we describe a machine learning approach, scTherapy, which leverages single-cell transcriptomic profiles to prioritize multi-targeting treatment options for individual patients with hematological cancers or solid tumors. Patient-specific treatments reveal a wide spectrum of co-inhibitors of multiple biological pathways predicted for primary cells from heterogenous cohorts of patients with acute myeloid leukemia and high-grade serous ovarian carcinoma, each with unique resistance patterns and synergy mechanisms. Experimental validations confirm that 96% of the multi-targeting treatments exhibit selective efficacy or synergy, and 83% demonstrate low toxicity to normal cells, highlighting their potential for therapeutic efficacy and safety. In a pan-cancer analysis across five cancer types, 25% of the predicted treatments are shared among the patients of the same tumor type, while 19% of the treatments are patient-specific. Our approach provides a widely-applicable strategy to identify personalized treatment regimens that selectively co-inhibit malignant cells and avoid inhibition of non-cancerous cells, thereby increasing their likelihood for clinical success.

Exploring single-cell RNA sequencing as a decision-making tool in the clinical management of Fuchs' endothelial corneal dystrophy

Single-cell RNA sequencing (scRNA-seq) has enabled the identification of novel gene signatures and cell heterogeneity in numerous tissues and diseases. Here we review the use of this technology for Fuchs' Endothelial Corneal Dystrophy (FECD). FECD is the most common indication for corneal endothelial transplantation worldwide. FECD is challenging to manage because it is genetically heterogenous, can be autosomal dominant or sporadic, and progress at different rates. Single-cell RNA sequencing has enabled the discovery of several FECD subtypes, each with associated gene signatures, and cell heterogeneity. Current FECD treatments are mainly surgical, with various Rho kinase (ROCK) inhibitors used to promote endothelial cell metabolism and proliferation following surgery. A range of emerging therapies for FECD including cell therapies, gene therapies, tissue engineered scaffolds, and pharmaceuticals are in preclinical and clinical trials. Unlike conventional disease management methods based on clinical presentations and family history, targeting FECD using scRNA-seq based precision-medicine has the potential to pinpoint the disease subtypes, mechanisms, stages, severities, and help clinicians in making the best decision for surgeries and the applications of therapeutics. In this review, we first discuss the feasibility and potential of using scRNA-seq in clinical diagnostics for FECD, highlight advances from the latest clinical treatments and emerging therapies for FECD, integrate scRNA-seq results and clinical notes from our FECD patients and discuss the potential of applying alternative therapies to manage these cases clinically.

Innovative aspects and applications of single cell technology for different diseases

Recent developments in single-cell technologies have provided valuable insights from cancer genomics to complex microbial communities. Single-cell technologies including the RNA-seq, next-generation sequencing (NGS), epigenomics, genomics, and transcriptomics can be used to uncover the single cell nature and molecular characterization of individual cells. These technologies also reveal the cellular transition states, evolutionary relationships between genes, the complex structure of single-cell populations, cell-to-cell interaction leading to biological discoveries and more reliable than traditional bulk technologies. These technologies are becoming the first choice for the early detection of inflammatory biomarkers affecting the proliferation and progression of tumor cells in the tumor microenvironment and improving the clinical efficacy of patients undergoing immunotherapy. These technologies also hold a central position in the detection of checkpoint inhibitors and thus determining the signaling pathways evoked by tumor invasion. This review addressed the emerging approaches of single cell-based technologies in cancer immunotherapies and different human diseases at cellular and molecular levels and the emerging role of sequencing technologies leading to drug discovery. Advancements in these technologies paved for discovering novel diagnostic markers for better understanding the pathological and biochemical mechanisms also for controlling the rate of different diseases.

scDrug+: predicting drug-responses using single-cell transcriptomics and molecular structure

Predicting drug responses based on individual transcriptomic profiles holds promise for refining prognosis and advancing precision medicine. Although many studies have endeavored to predict the responses of known drugs to novel transcriptomic profiles, research into predicting responses for newly discovered drugs remains sparse. In this study, we introduce scDrug+, a comprehensive pipeline that seamlessly integrates single-cell analysis with drug-response prediction. Importantly, scDrug+ is equipped to predict the response of new drugs by analyzing their molecular structures. The open-source tool is available as a Docker container, ensuring ease of deployment and reproducibility. It can be accessed at https://github.com/ailabstw/scDrugplus.

OmicVerse: a framework for bridging and deepening insights across bulk and single-cell sequencing

Single-cell sequencing is frequently affected by "omission" due to limitations in sequencing throughput, yet bulk RNA-seq may contain these ostensibly "omitted" cells. Here, we introduce the single cell trajectory blending from Bulk RNA-seq (BulkTrajBlend) algorithm, a component of the OmicVerse suite that leverages a Beta-Variational AutoEncoder for data deconvolution and graph neural networks for the discovery of overlapping communities. This approach effectively interpolates and restores the continuity of "omitted" cells within single-cell RNA sequencing datasets. Furthermore, OmicVerse provides an extensive toolkit for both bulk and single cell RNA-seq analysis, offering seamless access to diverse methodologies, streamlining computational processes, fostering exquisite data visualization, and facilitating the extraction of significant biological insights to advance scientific research.

Assessment of corneal nerve regeneration after axotomy in a compartmentalized microfluidic chip model with automated 3D high resolution live-imaging

Introduction

Damage to the corneal nerves can result in discomfort and chronic pain, profoundly impacting the quality of life of patients. Development of novel in vitro method is crucial to better understand corneal nerve regeneration and to find new treatments for the patients. Existing in vitro models often overlook the physiology of primary sensory neurons, for which the soma is separated from the nerve endings.

Methods

To overcome this limitation, our novel model combines a compartmentalized microfluidic culture of trigeminal ganglion neurons from adult mice with live-imaging and automated 3D image analysis offering robust way to assess axonal regrowth after axotomy.

Results

Physical axotomy performed by a two-second aspiration led to a reproducible 70% axonal loss and altered the phenotype of the neurons, increasing the number of substance P-positive neurons 72 h post-axotomy. To validate our new model, we investigated axonal regeneration after exposure to pharmacological compounds. We selected various targets known to enhance or inhibit axonal regrowth and analyzed their basal expression in trigeminal ganglion cells by scRNAseq. NGF/GDNF, insulin, and Dooku-1 (Piezo1 antagonist) enhanced regrowth by 81, 74 and 157%, respectively, while Yoda-1 (Piezo1 agonist) had no effect. Furthermore, SARM1-IN-2 (Sarm1 inhibitor) inhibited axonal regrowth, leading to only 6% regrowth after 72 h of exposure (versus 34% regrowth without any compound).

Discussion

Combining compartmentalized trigeminal neuronal culture with advanced imaging and analysis allowed a thorough evaluation of the extent of the axotomy and subsequent axonal regrowth. This innovative approach holds great promise for advancing our understanding of corneal nerve injuries and regeneration and ultimately improving the quality of life for patients suffering from sensory abnormalities, and related conditions.

Integrative analysis with machine learning identifies diagnostic and prognostic signatures in neuroblastoma based on differentially DNA methylated enhancers between INSS stage 4 and 4S neuroblastoma

INTRODUCTION

Accumulating evidence demonstrates that aberrant methylation of enhancers is crucial in gene expression profiles across several cancers. However, the latent effect of differently expressed enhancers between INSS stage 4S and 4 neuroblastoma (NB) remains elusive.

METHODS

We utilized the transcriptome and methylation data of stage 4S and 4 NB patients to perform Enhancer Linking by Methylation/Expression Relationships (ELMER) analysis, discovering a differently expressed motif within 67 enhancers between stage 4S and 4 NB. Harnessing the 67 motif genes, we established the INSS stage related signature (ISRS) by amalgamating 12 and 10 distinct machine learning (ML) algorithms across 113 and 101 ML combinations to precisely diagnose stage 4 NB among all NB patients and to predict the prognosis of NB patients. Based on risk scores calculated by prognostic ISRS, patients were categorized into high and low-risk groups according to median risk score. We conducted comprehensive comparisons between two risk groups, in terms of clinical applications, immune microenvironment, somatic mutations, immunotherapy, chemotherapy and single-cell analysis. Ultimately, we empirically validated the differential expressions of two ISRS model genes, CAMTA2 and FOXD1, through immunochemistry staining.

RESULTS

Through leave-one-out cross-validation, in both feature selection and model construction, we selected the random forest algorithm to diagnose stage 4 NB, and Enet algorithm to develop prognostic ISRS, due to their highest average C-index across five NB cohorts. After validations, the ISRS demonstrated a stable predictive capability, outperforming the previously published NB signatures and several clinic variables. We stratified NB patients into high and low-risk group based on median risk score, which showed the low-risk group with a superior survival outcome, an abundant immune infiltration, a decreased mutation landscape, and an enhanced sensitivity to immunotherapy. Single-cell analysis between two risk groups reveals biologically cellular variations underlying ISRS. Finally, we verified the significantly higher protein levels of CAMTA2 and FOXD1 in stage 4S NB, as well as their protective prognosis value in NB.

CONCLUSION

Based on multi-omics data and ML algorithms, we successfully developed the ISRS to enable accurate diagnosis and prognostic stratification in NB, which shed light on molecular mechanisms of spontaneous regression and clinical utilization of ISRS.

scDrugPrio: a framework for the analysis of single-cell transcriptomics to address multiple problems in precision medicine in immune-mediated inflammatory diseases

BACKGROUND

Ineffective drug treatment is a major problem for many patients with immune-mediated inflammatory diseases (IMIDs). Important reasons are the lack of systematic solutions for drug prioritisation and repurposing based on characterisation of the complex and heterogeneous cellular and molecular changes in IMIDs.

METHODS

Here, we propose a computational framework, scDrugPrio, which constructs network models of inflammatory disease based on single-cell RNA sequencing (scRNA-seq) data. scDrugPrio constructs detailed network models of inflammatory diseases that integrate information on cell type-specific expression changes, altered cellular crosstalk and pharmacological properties for the selection and ranking of thousands of drugs.

RESULTS

scDrugPrio was developed using a mouse model of antigen-induced arthritis and validated by improved precision/recall for approved drugs, as well as extensive in vitro, in vivo, and in silico studies of drugs that were predicted, but not approved, for the studied diseases. Next, scDrugPrio was applied to multiple sclerosis, Crohn's disease, and psoriatic arthritis, further supporting scDrugPrio through prioritisation of relevant and approved drugs. However, in contrast to the mouse model of arthritis, great interindividual cellular and gene expression differences were found in patients with the same diagnosis. Such differences could explain why some patients did or did not respond to treatment. This explanation was supported by the application of scDrugPrio to scRNA-seq data from eleven individual Crohn's disease patients. The analysis showed great variations in drug predictions between patients, for example, assigning a high rank to anti-TNF treatment in a responder and a low rank in a nonresponder to that treatment.

CONCLUSIONS

We propose a computational framework, scDrugPrio, for drug prioritisation based on scRNA-seq of IMID disease. Application to individual patients indicates scDrugPrio's potential for personalised network-based drug screening on cellulome-, genome-, and drugome-wide scales. For this purpose, we made scDrugPrio into an easy-to-use R package ( https://github.com/SDTC-CPMed/scDrugPrio ).

A review of computational methods for predicting cancer drug response at the single-cell level through integration with bulk RNAseq data

Cancer treatment failure is often attributed to tumor heterogeneity, where diverse malignant cell clones exist within a patient. Despite a growing understanding of heterogeneous tumor cells depicted by single-cell RNA sequencing (scRNA-seq), there is still a gap in the translation of such knowledge into treatment strategies tackling the pervasive issue of therapy resistance. In this review, we survey methods leveraging large-scale drug screens to generate cellular sensitivities to various therapeutics. These methods enable efficient drug screens in scRNA-seq data and serve as the bedrock of drug discovery for specific cancer cell groups. We envision that they will become an indispensable tool for tailoring patient care in the era of heterogeneity-aware precision medicine.

Single-cell transcriptomic analysis reveals tumor cell heterogeneity and immune microenvironment features of pituitary neuroendocrine tumors

BACKGROUND

Pituitary neuroendocrine tumors (PitNETs) are one of the most common types of intracranial tumors. Currently, the cellular characteristics of normal pituitary and various other types of PitNETs are still not completely understood.

METHODS

We performed single-cell RNA sequencing (scRNA-seq) on 4 normal samples and 24 PitNET samples for comprehensive bioinformatics analysis. Findings regarding the function of PBK in the aggressive tumor cells were validated by siRNA knockdown, overexpression, and transwell experiments.

RESULTS

We first constructed a reference cell atlas of the human pituitary. Subsequent scRNA-seq analysis of PitNET samples, representing major tumor subtypes, shed light on the intrinsic cellular heterogeneities of the tumor cells and tumor microenvironment (TME). We found that the expression of hormone-encoding genes defined the major variations of the PIT1-lineage tumor cell transcriptomic heterogeneities. A sub-population of TPIT-lineage tumor cells highly expressing GZMK suggested a novel subtype of corticotroph tumors. In immune cells, we found two clusters of tumor-associated macrophages, which were both highly enriched in PitNETs but with distinct functional characteristics. In PitNETs, the stress response pathway was significantly activated in T cells. While a majority of these tumors are benign, our study unveils a common existence of aggressive tumor cells in the studied samples, which highly express a set of malignant signature genes. The following functional experiments confirmed the oncogenic role of selected up-regulated genes. The over-expression of PBK could promote both tumor cell proliferation and migration, and it was also significantly associated with poor prognosis in PitNET patients.

CONCLUSIONS

Our data and analysis manifested the basic cell types in the normal pituitary and inherent heterogeneity of PitNETs, identified several features of the tumor immune microenvironments, and found a novel epithelial cell sub-population with aggressive signatures across all the studied cases.

Predicting drug response from single-cell expression profiles of tumours

BACKGROUND

Intra-tumour heterogeneity (ITH) presents a significant obstacle in formulating effective treatment strategies in clinical practice. Single-cell RNA sequencing (scRNA-seq) has evolved as a powerful instrument for probing ITH at the transcriptional level, offering an unparalleled opportunity for therapeutic intervention.

RESULTS

Drug response prediction at the single-cell level is an emerging field of research that aims to improve the efficacy and precision of cancer treatments. Here, we introduce DREEP (Drug Response Estimation from single-cell Expression Profiles), a computational method that leverages publicly available pharmacogenomic screens from GDSC2, CTRP2, and PRISM and functional enrichment analysis to predict single-cell drug sensitivity from transcriptomic data. We validated DREEP extensively in vitro using several independent single-cell datasets with over 200 cancer cell lines and showed its accuracy and robustness. Additionally, we also applied DREEP to molecularly barcoded breast cancer cells and identified drugs that can selectively target specific cell populations.

CONCLUSIONS

DREEP provides an in silico framework to prioritize drugs from single-cell transcriptional profiles of tumours and thus helps in designing personalized treatment strategies and accelerating drug repurposing studies. DREEP is available at https://github.com/gambalab/DREEP .

scDrugPrio: A framework for the analysis of single-cell transcriptomics to address multiple problems in precision medicine in immune-mediated inflammatory diseases

Background

Ineffective drug treatment is a major problem for many patients with immune-mediated inflammatory diseases (IMIDs). Important reasons are the lack of systematic solutions for drug prioritisation and repurposing based on characterisation of the complex and heterogeneous cellular and molecular changes in IMIDs.

Methods

Here, we propose a computational framework, scDrugPrio, which constructs network models of inflammatory disease based on single-cell RNA sequencing (scRNA-seq) data. scDrugPrio constructs detailed network models of inflammatory diseases that integrate information on cell type-specific expression changes, altered cellular crosstalk and pharmacological properties for the selection and ranking of thousands of drugs.

Results

scDrugPrio was developed using a mouse model of antigen-induced arthritis and validated by improved precision/recall for approved drugs, as well as extensive in vitro, in vivo, and in silico studies of drugs that were predicted, but not approved, for the studied diseases. Next, scDrugPrio was applied to multiple sclerosis, Crohn's disease, and psoriatic arthritis, further supporting scDrugPrio through prioritisation of relevant and approved drugs. However, in contrast to the mouse model of arthritis, great interindividual cellular and gene expression differences were found in patients with the same diagnosis. Such differences could explain why some patients did or did not respond to treatment. This explanation was supported by the application of scDrugPrio to scRNA-seq data from eleven individual Crohn's disease patients. The analysis showed great variations in drug predictions between patients, for example, assigning a high rank to anti-TNF treatment in a responder and a low rank in a nonresponder to that treatment.

Conclusion

We propose a computational framework, scDrugPrio, for drug prioritisation based on scRNA-seq of IMID disease. Application to individual patients indicates scDrugPrio's potential for personalised network-based drug screening on cellulome-, genome-, and drugome-wide scales. For this purpose, we made scDrugPrio into an easy-to-use R package (https://github.com/SDTC-CPMed/scDrugPrio).

scMultiGAN: cell-specific imputation for single-cell transcriptomes with multiple deep generative adversarial networks

The emergence of single-cell RNA sequencing (scRNA-seq) technology has revolutionized the identification of cell types and the study of cellular states at a single-cell level. Despite its significant potential, scRNA-seq data analysis is plagued by the issue of missing values. Many existing imputation methods rely on simplistic data distribution assumptions while ignoring the intrinsic gene expression distribution specific to cells. This work presents a novel deep-learning model, named scMultiGAN, for scRNA-seq imputation, which utilizes multiple collaborative generative adversarial networks (GAN). Unlike traditional GAN-based imputation methods that generate missing values based on random noises, scMultiGAN employs a two-stage training process and utilizes multiple GANs to achieve cell-specific imputation. Experimental results show the efficacy of scMultiGAN in imputation accuracy, cell clustering, differential gene expression analysis and trajectory analysis, significantly outperforming existing state-of-the-art techniques. Additionally, scMultiGAN is scalable to large scRNA-seq datasets and consistently performs well across sequencing platforms. The scMultiGAN code is freely available at https://github.com/Galaxy8172/scMultiGAN.