iGWAS: Image-based genome-wide association of self-supervised deep phenotyping of retina fundus images.

Existing imaging genetics studies have been mostly limited in scope by using imaging-derived phenotypes defined by human experts. Here, leveraging new breakthroughs in self-supervised deep representation learning, we propose a new approach, image-based genome-wide association study (iGWAS), for identifying genetic factors associated with phenotypes discovered from medical images using contrastive learning. Using retinal fundus photos, our model extracts a 128-dimensional vector representing features of the retina as phenotypes. After training the model on 40,000 images from the EyePACS dataset, we generated phenotypes from 130,329 images of 65,629 British White participants in the UK Biobank. We conducted GWAS on these phenotypes and identified 14 loci with genome-wide significance (p<5×10-8 and intersection of hits from left and right eyes). We also did GWAS on the retina color, the average color of the center region of the retinal fundus photos. The GWAS of retina colors identified 34 loci, 7 are overlapping with GWAS of raw image phenotype. Our results establish the feasibility of this new framework of genomic study based on self-supervised phenotyping of medical images.

To be or not to be - Decoding the Trabecular Meshwork Cell Identity.

The trabecular meshwork within the conventional outflow apparatus is critical in maintaining intraocular pressure homeostasis. In vitro studies employing primary cell cultures of the human trabecular meshwork (hTM) have conventionally served as surrogates for investigating the pathobiology of TM dysfunction. Despite its abundant use, translation of outcomes from in vitro studies to ex vivo and/or in vivo studies remains a challenge. Given the cell heterogeneity, performing single-cell RNA sequencing comparing primary hTM cell cultures to hTM tissue may provide important insights on cellular identity and translatability, as such an approach has not been reported before. In this study, we assembled a total of 14 primary hTM in vitro samples across passages 1-4, including 4 samples from individuals diagnosed with glaucoma. This dataset offers a comprehensive transcriptomic resource of primary hTM in vitro scRNA-seq data to study global changes in gene expression in comparison to cells in tissue in situ. We have performed extensive preprocessing and quality control, allowing the research community to access and utilize this public resource.

Observations of the delayed-choice quantum eraser using coherent photons.

Quantum superposition is the cornerstone of quantum mechanics, where interference fringes originate in the self-interference of a single photon via indistinguishable photon characteristics. Wheeler's delayed-choice experiments have been extensively studied for the wave-particle duality over the last several decades to understand the complementarity theory of quantum mechanics. The heart of the delayed-choice quantum eraser is in the mutually exclusive quantum feature violating the cause-effect relation. Here, we experimentally demonstrate the quantum eraser using coherent photon pairs by the delayed choice of a polarizer placed out of the interferometer. Coherence solutions of the observed quantum eraser are derived from a typical Mach-Zehnder interferometer, where the violation of the cause-effect relation is due to selective measurements of basis choice.

SOX7 deficiency causes ventricular septal defects through its effects on endocardial-to-mesenchymal transition and the expression of Wnt4 and Bmp2.

SOX7 is a transcription factor-encoding gene located in a region on chromosome 8p23.1 that is recurrently deleted in individuals with ventricular septal defects (VSDs). We have previously shown that Sox7-/- embryos die of heart failure around E11.5. Here, we demonstrate that these embryos have hypocellular endocardial cushions with severely reduced numbers of mesenchymal cells. Ablation of Sox7 in the endocardium also resulted in hypocellular endocardial cushions, and we observed VSDs in rare E15.5 Sox7flox/-;Tie2-Cre and Sox7flox/flox;Tie2-Cre embryos that survived to E15.5. In atrioventricular explant studies, we showed that SOX7 deficiency leads to a severe reduction in endocardial-to-mesenchymal transition (EndMT). RNA-seq studies performed on E9.5 Sox7-/- heart tubes revealed severely reduced Wnt4 transcript levels. Wnt4 is expressed in the endocardium and promotes EndMT by acting in a paracrine manner to increase the expression of Bmp2 in the myocardium. Both WNT4 and BMP2 have been previously implicated in the development of VSDs in individuals with 46,XX sex reversal with dysgenesis of kidney, adrenals and lungs (SERKAL) syndrome and in individuals with short stature, facial dysmorphism and skeletal anomalies with or without cardiac anomalies 1 (SSFSC1) syndrome, respectively. We now show that Sox7 and Wnt4 interact genetically in the development of VSDs through their additive effects on endocardial cushion development with Sox7+/-;Wnt4+/- double heterozygous embryos having hypocellular endocardial cushions and perimembranous and muscular VSDs not seen in their Sox7+/- and Wnt4+/- littermates. These results provide additional evidence that SOX7, WNT4 and BMP2 function in the same pathway during mammalian septal development and that their deficiency can contribute to the development of VSDs in humans.

A Spontaneous Nonhuman Primate Model of Myopic Foveoschisis.

Purpose: Foveoschisis involves the pathologic splitting of retinal layers at the fovea, which may occur congenitally in X-linked retinoschisis (XLRS) or as an acquired complication of myopia. XLRS is attributed to functional loss of the retinal adhesion protein retinoschisin 1 (RS1), but the pathophysiology of myopic foveoschisis is unclear due to the lack of animal models. Here, we characterized a novel nonhuman primate model of myopic foveoschisis through clinical examination and multimodal imaging followed by morphologic, cellular, and transcriptional profiling of retinal tissues and genetic analysis. Methods: We identified a rhesus macaque with behavioral and anatomic features of myopic foveoschisis, and monitored disease progression over 14 months by fundus photography, fluorescein angiography, and optical coherence tomography (OCT). After necropsy, we evaluated anatomic and cellular changes by immunohistochemistry and transcriptomic changes using single-nuclei RNA-sequencing (snRNA-seq). Finally, we performed Sanger and whole exome sequencing with focus on the RS1 gene. Results: Affected eyes demonstrated posterior hyaloid traction and progressive splitting of the outer plexiform layer on OCT. Immunohistochemistry showed increased GFAP expression in Müller glia and loss of ramified Iba-1+ microglia, suggesting macro- and microglial activation with minimal photoreceptor alterations. SnRNA-seq revealed gene expression changes predominantly in cones and retinal ganglion cells involving chromatin modification, suggestive of cellular stress at the fovea. No defects in the RS1 gene or its expression were detected. Conclusions: This nonhuman primate model of foveoschisis reveals insights into how acquired myopic traction leads to phenotypically similar morphologic and cellular changes as congenital XLRS without alterations in RS1.

Revisiting self-interference in Young's double-slit experiments.

Quantum superposition is the heart of quantum mechanics as mentioned by Dirac and Feynman. In an interferometric system, single photon self-interference has been intensively studied over the last several decades in both quantum and classical regimes. In Born rule tests, the Sorkin parameter indicates the maximum number of possible quantum superposition allowed to the input photons entering an interferometer, where multi-photon interference fringe is equivalent to that of a classical version by a laser. Here, an attenuated laser light in a quantum regime is investigated for self-interference in a Mach-Zehnder interferometer, and the results are compared with its classical version. The equivalent result supports the Born rule tests, where the classical interference originates in the superposition of individual single-photon self-interferences. This understanding sheds light on the fundamental physics of quantum features between bipartite systems.

Bi-order multimodal integration of single-cell data.

Integration of single-cell multiomics profiles generated by different single-cell technologies from the same biological sample is still challenging. Previous approaches based on shared features have only provided approximate solutions. Here, we present a novel mathematical solution named bi-order canonical correlation analysis (bi-CCA), which extends the widely used CCA approach to iteratively align the rows and the columns between data matrices. Bi-CCA is generally applicable to combinations of any two single-cell modalities. Validations using co-assayed ground truth data and application to a CAR-NK study and a fetal muscle atlas demonstrate its capability in generating accurate multimodal co-embeddings and discovering cellular identity.

Assessing Variant Causality and Severity Using Retinal Pigment Epithelial Cells Derived from Stargardt Disease Patients.

Purpose: Modern molecular genetics has revolutionized gene discovery, genetic diagnoses, and precision medicine yet many patients remain unable to benefit from these advances as disease-causing variants remain elusive for up to half of Mendelian genetic disorders. Patient-derived induced pluripotent stem (iPS) cells and transcriptomics were used to identify the fate of unsolved ABCA4 alleles in patients with Stargardt disease. Methods: Multiple independent iPS lines were generated from skin biopsies of three patients with Stargardt disease harboring a single identified pathogenic ABCA4 variant. Derived retinal pigment epithelial cells (dRPE) from a normal control and patient cells were subjected to RNA-Seq on the Novaseq6000 platform, analyzed using DESeq2 with calculation of allele specific imbalance from the pathogenic or a known linked variant. Protein analysis was performed using the automated Simple Western system. Results: Nine dRPE samples were generated, with transcriptome analysis on eight. Allele-specific expression indicated normal transcripts expressed from splice variants albeit at low levels, and missense transcripts expressed at near-normal levels. Corresponding protein was not easily detected. Patient phenotype correlation indicated missense variants expressed at high levels have more deleterious outcomes. Transcriptome analysis suggests mitochondrial membrane biodynamics and the unfolded protein response pathway may be relevant in Stargardt disease. Conclusions: Patient-specific iPS-derived RPE cells set the stage to assess non-expressing variants in difficult-to-detect genomic regions using easily biopsied tissue. Translational Relevance: This "Disease in a Dish" approach is likely to enhance the ability of patients to participate in and benefit from clinical trials while providing insights into perturbations in RPE biology.

Retinal organoids derived from rhesus macaque iPSCs undergo accelerated differentiation compared to human stem cells.

PURPOSE: To compare the timing and efficiency of the development of Macaca mulatta, a nonhuman primate (NHP), induced pluripotent stem cell (rhiPSC) derived retinal organoids to those derived from human embryonic stem cells (hESCs). RESULTS: Generation of retinal organoids was achieved from both human and several NHP pluripotent stem cell lines. All rhiPSC lines resulted in retinal differentiation with the formation of optic vesicle-like structures similar to what has been observed in hESC retinal organoids. NHP retinal organoids had laminated structure and were composed of mature retinal cell types including cone and rod photoreceptors. Single-cell RNA sequencing was conducted at two time points; this allowed identification of cell types and developmental trajectory characterization of the developing organoids. Important differences between rhesus and human cells were measured regarding the timing and efficiency of retinal organoid differentiation. While the culture of NHP-derived iPSCs is relatively difficult compared to that of human stem cells, the generation of retinal organoids from NHP iPSCs is feasible and may be less time-consuming due to an intrinsically faster timing of retinal differentiation. CONCLUSIONS: Retinal organoids produced from rhesus monkey iPSCs using established protocols differentiate through the stages of organoid development faster than those derived from human stem cells. The production of NHP retinal organoids may be advantageous to reduce experimental time for basic biology studies in retinogenesis as well as for preclinical trials in NHPs studying retinal allograft transplantation.

Single-Cell Transcriptomics Identifies a Unique Entity and Signature Markers of Transit-Amplifying Cells in Human Corneal Limbus.

Purpose: Differentiated from adult stem cells (ASCs), transit-amplifying cells (TACs) play an important role in tissue homeostasis, development, and regeneration. This study aimed to characterize the gene expression profile of a candidate TAC population in limbal basal epithelial cells using single-cell RNA sequencing (scRNA-seq). Methods: Single cells isolated from the basal corneal limbus were subjected to scRNA-seq using the 10x Genomics platform. Cell types were clustered by graph-based visualization methods and unbiased computational analysis. BrdU proliferation assays, immunofluorescent staining, and real-time reverse transcription quantitative polymerase chain reaction were performed using multiple culture models of primary human limbal epithelial cells to characterize the TAC pool. Results: Single-cell transcriptomics of 16,360 limbal basal cells revealed 12 cell clusters. A unique cluster (3.21% of total cells) was identified as a TAC entity, based on its less differentiated progenitor status and enriched exclusive proliferation marker genes, with 98.1% cells in S and G2/M phases. The cell cycle-dependent genes were revealed to be largely enriched by the TAC population. The top genes were characterized morphologically and functionally at protein and mRNA levels. The specific expression patterns of RRM2, TK1, CENPF, NUSAP1, UBE2C, and CDC20 were well correlated in a time- and cycle-dependent manner with proliferation stages in the cell growth and regeneration models. Conclusions: For the first time, to the best of our knowledge, we have identified a unique TAC entity and uncovered a group of cell cycle-dependent genes that serve as TAC signature markers. The findings provide insight into ASCs and TACs and lay the foundation for understanding corneal homeostasis and diseases.

Single-cell transcriptomics identifies limbal stem cell population and cell types mapping its differentiation trajectory in limbal basal epithelium of human cornea.

PURPOSE: This study aimed to uncover novel cell types in heterogenous basal limbus of human cornea for identifying LSC at single cell resolution. METHODS: Single cells of human limbal basal epithelium were isolated from young donor corneas. Single-cell RNA-Sequencing was performed using 10x Genomics platform, followed by clustering cell types through the graph-based visualization method UMAP and unbiased computational informatic analysis. Tissue RNA in situ hybridization with RNAscope, immunofluorescent staining and multiple functional assays were performed using human corneas and limbal epithelial culture models. RESULTS: Single-cell transcriptomics of 16,360 limbal basal cells revealed 12 cell clusters belonging to three lineages. A smallest cluster (0.4% of total cells) was identified as LSCs based on their quiescent and undifferentiated states with enriched marker genes for putative epithelial stem cells. TSPAN7 and SOX17 are discovered and validated as new LSC markers based on their exclusive expression pattern and spatial localization in limbal basal epithelium by RNAscope and immunostaining, and functional role in cell growth and tissue regeneration models with RNA interference in cultures. Interestingly, five cell types/states mapping a developmental trajectory of LSC from quiescence to proliferation and differentiation are uncovered by Monocle3 and CytoTRACE pseudotime analysis. The transcription factor networks linking novel signaling pathways are revealed to maintain LSC stemness. CONCLUSIONS: This human corneal scRNA-Seq identifies the LSC population and uncovers novel cell types mapping the differentiation trajectory in heterogenous limbal basal epithelium. The findings provide insight into LSC concept and lay the foundation for understanding the corneal homeostasis and diseases.

Single-Cell Transcriptome Analysis Reveals Dynamic Cell Populations and Differential Gene Expression Patterns in Control and Aneurysmal Human Aortic Tissue.

BACKGROUND: Ascending thoracic aortic aneurysm (ATAA) is caused by the progressive weakening and dilatation of the aortic wall and can lead to aortic dissection, rupture, and other life-threatening complications. To improve our understanding of ATAA pathogenesis, we aimed to comprehensively characterize the cellular composition of the ascending aortic wall and to identify molecular alterations in each cell population of human ATAA tissues. METHODS: We performed single-cell RNA sequencing analysis of ascending aortic tissues from 11 study participants, including 8 patients with ATAA (4 women and 4 men) and 3 control subjects (2 women and 1 man). Cells extracted from aortic tissue were analyzed and categorized with single-cell RNA sequencing data to perform cluster identification. ATAA-related changes were then examined by comparing the proportions of each cell type and the gene expression profiles between ATAA and control tissues. We also examined which genes may be critical for ATAA by performing the integrative analysis of our single-cell RNA sequencing data with publicly available data from genome-wide association studies. RESULTS: We identified 11 major cell types in human ascending aortic tissue; the high-resolution reclustering of these cells further divided them into 40 subtypes. Multiple subtypes were observed for smooth muscle cells, macrophages, and T lymphocytes, suggesting that these cells have multiple functional populations in the aortic wall. In general, ATAA tissues had fewer nonimmune cells and more immune cells, especially T lymphocytes, than control tissues did. Differential gene expression data suggested the presence of extensive mitochondrial dysfunction in ATAA tissues. In addition, integrative analysis of our single-cell RNA sequencing data with public genome-wide association study data and promoter capture Hi-C data suggested that the erythroblast transformation-specific related gene(ERG) exerts an important role in maintaining normal aortic wall function. CONCLUSIONS: Our study provides a comprehensive evaluation of the cellular composition of the ascending aortic wall and reveals how the gene expression landscape is altered in human ATAA tissue. The information from this study makes important contributions to our understanding of ATAA formation and progression.

Single-Cell Capture, RNA-seq, and Transcriptome Analysis from the Neural Retina.

Single-cell RNA sequencing (scRNA-seq) is an emerging technology that can address the challenge of cellular heterogeneity. In the last decade, the cost per cell has been dramatically reduced, and the throughput has been increased by 104-fold. Like many other tissues, the retina is highly heterogeneous with an estimated of over 100 subtypes of neuronal cells. Here, we describe the current techniques to perform scRNA-seq on the adult retinal tissue including retinal dissection, retinal dissociation, assessment of cell population, cDNA synthesis, library construction, and next-generation sequencing. In addition, we introduce a workflow of scRNA-seq data analysis using open-source tools.

Single-nuclei RNA-seq on human retinal tissue provides improved transcriptome profiling.

Single-cell RNA-seq is a powerful tool in decoding the heterogeneity in complex tissues by generating transcriptomic profiles of the individual cell. Here, we report a single-nuclei RNA-seq (snRNA-seq) transcriptomic study on human retinal tissue, which is composed of multiple cell types with distinct functions. Six samples from three healthy donors are profiled and high-quality RNA-seq data is obtained for 5873 single nuclei. All major retinal cell types are observed and marker genes for each cell type are identified. The gene expression of the macular and peripheral retina is compared to each other at cell-type level. Furthermore, our dataset shows an improved power for prioritizing genes associated with human retinal diseases compared to both mouse single-cell RNA-seq and human bulk RNA-seq results. In conclusion, we demonstrate that obtaining single cell transcriptomes from human frozen tissues can provide insight missed by either human bulk RNA-seq or animal models.

Generation, transcriptome profiling, and functional validation of cone-rich human retinal organoids.

Rod and cone photoreceptors are light-sensing cells in the human retina. Rods are dominant in the peripheral retina, whereas cones are enriched in the macula, which is responsible for central vision and visual acuity. Macular degenerations affect vision the most and are currently incurable. Here we report the generation, transcriptome profiling, and functional validation of cone-rich human retinal organoids differentiated from hESCs using an improved retinal differentiation system. Induced by extracellular matrix, aggregates of hESCs formed single-lumen cysts composed of epithelial cells with anterior neuroectodermal/ectodermal fates, including retinal cell fate. Then, the cysts were en bloc-passaged, attached to culture surface, and grew, forming colonies in which retinal progenitor cell patches were found. Following gentle cell detachment, retinal progenitor cells self-assembled into retinal epithelium-retinal organoid-that differentiated into stratified cone-rich retinal tissue in agitated cultures. Electron microscopy revealed differentiating outer segments of photoreceptor cells. Bulk RNA-sequencing profiling of time-course retinal organoids demonstrated that retinal differentiation in vitro recapitulated in vivo retinogenesis in temporal expression of cell differentiation markers and retinal disease genes, as well as in mRNA alternative splicing. Single-cell RNA-sequencing profiling of 8-mo retinal organoids identified cone and rod cell clusters and confirmed the cone enrichment initially revealed by quantitative microscopy. Notably, cones from retinal organoids and human macula had similar single-cell transcriptomes, and so did rods. Cones in retinal organoids exhibited electrophysiological functions. Collectively, we have established cone-rich retinal organoids and a reference of transcriptomes that are valuable resources for retinal studies.

Transcriptomic profiles of retinal ganglion cells are defined by the magnitude of intraocular pressure elevation in adult mice.

Elevated intraocular pressure (IOP) is the major risk factor for glaucoma, a sight threatening disease of retinal ganglion cells (RGCs) and their axons. Despite the central importance of IOP, details of the impact of IOP elevation on RGC gene expression remain elusive. We developed a 4-step immunopanning protocol to extract adult mouse RGCs with high fidelity and used it to isolate RGCs from wild type mice exposed to 2 weeks of IOP elevation generated by the microbead model. IOP was elevated to 2 distinct levels which were defined as Mild (IOP increase >1 mmHg and <4 mmHg) and Moderate (IOP increase ≥4 mmHg). RNA sequencing was used to compare the transcriptional environment at each IOP level. Differentially expressed genes were markedly different between the 2 groups, and pathway analysis revealed frequently opposed responses between the IOP levels. These results suggest that the magnitude of IOP elevation has a critical impact on RGC transcriptional changes. Furthermore, it is possible that IOP-based set points exist within RGCs to impact the direction of transcriptional change. It is possible that this improved understanding of changes in RGC gene expression can ultimately lead to novel diagnostics and therapeutics for glaucoma.

Dynamic locomotion synchronization of bipedal robot and human operator via bilateral feedback teleoperation.

Despite remarkable progress in artificial intelligence, autonomous humanoid robots are still far from matching human-level manipulation and locomotion proficiency in real applications. Proficient robots would be ideal first responders to dangerous scenarios such as natural or man-made disasters. When handling these situations, robots must be capable of navigating highly unstructured terrain and dexterously interacting with objects designed for human workers. To create humanoid machines with human-level motor skills, in this work, we use whole-body teleoperation to leverage human control intelligence to command the locomotion of a bipedal robot. The challenge of this strategy lies in properly mapping human body motion to the machine while simultaneously informing the operator how closely the robot is reproducing the movement. Therefore, we propose a solution for this bilateral feedback policy to control a bipedal robot to take steps, jump, and walk in synchrony with a human operator. Such dynamic synchronization was achieved by (i) scaling the core components of human locomotion data to robot proportions in real time and (ii) applying feedback forces to the operator that are proportional to the relative velocity between human and robot. Human motion was sped up to match a faster robot, or drag was generated to synchronize the operator with a slower robot. Here, we focused on the frontal plane dynamics and stabilized the robot in the sagittal plane using an external gantry. These results represent a fundamental solution to seamlessly combine human innate motor control proficiency with the physical endurance and strength of humanoid robots.

Integrated Genomic Comparison of Mouse Models Reveals Their Clinical Resemblance to Human Liver Cancer.

Hepatocellular carcinoma (HCC) is a heterogeneous disease. Mouse models are commonly used as preclinical models to study hepatocarcinogenesis, but how well these models recapitulate molecular subtypes of human HCC is unclear. Here, integration of genomic signatures from molecularly and clinically defined human HCC (n = 11) and mouse models of HCC (n = 9) identified the mouse models that best resembled subtypes of human HCC and determined the clinical relevance of each model. Mst1/2 knockout (KO), Sav1 KO, and SV40 T antigen mouse models effectively recapitulated subtypes of human HCC with a poor prognosis, whereas the Myc transgenic model best resembled human HCCs with a more favorable prognosis. The Myc model was also associated with activation of ß-catenin. E2f1, E2f1/Myc, E2f1/Tgfa, and diethylnitrosamine (DEN)-induced models were heterogeneous and were unequally split into poor and favorable prognoses. Mst1/2 KO and Sav1 KO models best resemble human HCC with hepatic stem cell characteristics. Applying a genomic predictor for immunotherapy, the six-gene IFNγ score, the Mst1/2 KO, Sav1 KO, SV40, and DEN models were predicted to be the least responsive to immunotherapy. Further analysis showed that elevated expression of immune-inhibitory genes (Cd276 and Nectin2/Pvrl2) in Mst1/2 KO, Sav1 KO, and SV40 models and decreased expression of immune stimulatory gene (Cd86) in the DEN model might be accountable for the lack of predictive response to immunotherapy.Implication: The current genomic approach identified the most relevant mouse models to human liver cancer and suggests immunotherapeutic potential for the treatment of specific subtypes. Mol Cancer Res; 16(11); 1713-23. ©2018 AACR.

Inhibition of pancreatic cancer Panc1 cell migration by omeprazole is dependent on aryl hydrocarbon receptor activation of JNK.

Several aryl hydrocarbon receptor (AhR)-active pharmaceuticals were screened as inhibitors of pancreatic cancer cell invasion and identified two compounds, omeprazole, that inhibited invasion. Inhibition of highly invasive Panc1 cell invasion by omeprazole involves an AhR-dependent non-genomic pathway, and omeprazole-mediated inhibition of Panc1 cell invasion was dependent on Jun-N-terminal kinase (JNK) and mitogen-activated kinase kinase 7 (MKK7). The failure of omeprazole to induce nuclear translocation of the AhR was not due to overexpression of cytosolic AhR partner proteins Hsp90 or XAP2, and results of DNA sequencing show that the AhR expressed in Panc1 cells was not mutated. Results of RNAseq studies indicate that omeprazole induced an AhR-dependent downregulation of several pro-invasion factors including activated leukocyte cell adhesion molecule (ALCAM), long chain fatty acid CoA-synthase (CSL4), stathmin 3 (STMN3) and neuropillin 2 (NRP2), and the specific functions of these genes are currently being investigated.

Clinical and genomic landscape of gastric cancer with a mesenchymal phenotype.

Gastric cancer is a heterogeneous cancer, making treatment responses difficult to predict. Here we show that we identify two distinct molecular subtypes, mesenchymal phenotype (MP) and epithelial phenotype (EP), by analyzing genomic and proteomic data. Molecularly, MP subtype tumors show high genomic integrity characterized by low mutation rates and microsatellite stability, whereas EP subtype tumors show low genomic integrity. Clinically, the MP subtype is associated with markedly poor survival and resistance to standard chemotherapy, whereas the EP subtype is associated with better survival rates and sensitivity to chemotherapy. Integrative analysis shows that signaling pathways driving epithelial-to-mesenchymal transition and insulin-like growth factor 1 (IGF1)/IGF1 receptor (IGF1R) pathway are highly activated in MP subtype tumors. Importantly, MP subtype cancer cells are more sensitive to inhibition of IGF1/IGF1R pathway than EP subtype. Detailed characterization of these two subtypes could identify novel therapeutic targets and useful biomarkers for prognosis and therapy response.